Anthropology: The Art of Building a Successful Social Site
Picture if you will, a collaborative site that runs on two servers, is managed by four people, and has attracted a third of its target demographic within six months of launch. A site that has had 800,000 posts submitted by its users in its short lifetime and has 16 million pageviews/month - and growing.
This is the story of Stack Overflow, a free question and answer site built by developers for developers that has fostered a strong and committed online community in under one year. How? Easy, according to founder Joel Spolsky; all it takes is an understanding of anthropology and a lot of determination.
"As we move from the era of computing into the era of the Internet, we no longer need to worry about computer-human interaction." Joel Spolsky told a group of programmers at Google last month. "What we do have to think about [in the era of social networking] is human to human interaction," he said. And according to Spolsky, to do that, you have to think as an anthropologist does.
Anthropology and the Social Web
"In anthropology it's very clear that the environment that you create influences people and how they behave", Spolsky explained. "People will come into the environment and behave according to what you built in certain subtle ways; ways that you probably didn't think about."
He points to the Scalinata della Trinita dei Monti, or the Spanish Steps to further his point. "They were built to be stairs," he said, "from the Spanish Embassy at the base, up to the Trinita dei Monti at the top." Instead, they've become the "living room" for backpackers in Rome. "Partially it has to do with the steps being the perfect comfortable height to sit on," he said, but also, they provide a fantastic view of the Piazza di Spagna at the bottom.
"This was completely non intentional," he explained. "Similarly, the user interface you create for your applications will influence how people behave."
So what is Stack Overflow and why Does it Matter?
Founded by Joel Spolsky and Jeff Atwood, Stack Overflow is a free question and answer site designed to help developers get the most relevant answers from peers - and fast. Collaboratively built and maintained by a legion of committed developers, it is OS and language agnostic.
Stack Overflow came about because search engines are failing in a particular realm said Spolsky: "Expert Q&A; where you can ask an expert and the expert can give you a true and correct answer." And while he points out that a lot of the companies organized around search have tried to make question and answer type portals; no one site provides value.
And according to Spolsky, there are a variety of reasons why these sites are ineffective: Yahoo! Answers attracts too many adolescents seeking answers to questions about "reproductive sciences;" Mahalo Answers, the brainchild of Jason Calacanis, creates an environment where questions appear to be "scams;" AskVille, Amazon's creation emphasizes the question rather than the answer with its oversize version of a search box. Of course, there is one more; the one that must not be named, and perhaps the unofficial Raison d'etre for Stack Overflow, is basically just a big sneaky tricky mess.
Why Search Engines are Failing when it Comes to Collaborative Sites
According to Spolsky, there are certain reasons why search engines are failing when it comes to Q&A sites, and they are the same issues Stack Overflow is trying to solve.
- Sign-up scams: Sites that a search engine may send you to where you must first sign up and pay, if you want an answer.
- Register: A "road bump" that many sites have, and one Spolsky thinks reduces participation dramatically
- Wrong answers: When searching for highly technical questions, a search engine may send you to a forum that has multiple answers. If you are unsure which answer is the correct one, you waste too much time working through the wrong ones.
- Obsolete results: Google, for instance, will oftentimes give an older page priority. In turn, the page you are served is often outdated and no longer relevant.
The Nine Building Blocks of Social Engineering
To work around these problems, Stack Overflow was built on what Spolsky calls the nine "building blocks" in an effort to create a site that was anthropologically correct and would encourage people to behave in a way that would work. He also pointed out that every single one is copied from somewhere else.
- Voting: Copied from Reddit, via Digg, voting allows people to vote up answers they think are good. Stack Overflow tweaked its voting algorithm, giving the person who asked the question special power to select one answer as the official answer that will rise to the top regardless of what the community voted. The second answer, of course is always the highest ranked community answer.
- Tags: Tags allow users to specify perspective. For instance, Spolsky explained, "you can add that I'm asking this from a VB perspective, not a C# perspective." Stack Overflow is also customizable with tags, allowing users to specify which technology they are interested in, and typical of most social sites. What is not typical however, is the ability to ignore tags that Stack Overflow has built in.
- Editing: Taking a page out of Wikipedia, Stack Overflow allows users to edit both questions and answers; so answers could get better, rather than becoming "this frozen artifact on the Internet until the end of time," which is typical of most forum threads.
- Badges: "A soldier will fight long and hard for a bit of colored ribbon," said Napoleon once upon a time, and so Stack Overflow made the decision to reward its users with badges. Over time, the badges show credibility.
- Karma: People are willing to do for free what they're not willing to do for small amounts of money according to Spolsky and by offering karma, Stack Overflow encourages its users to do more. More Karma equals more privileges on the site.
- Pre-search: Once you begin typing your question, Stack Overflow's pre-search will do a quick search to see whether the question has been asked before and display the result for easy access and to prevent duplication issues.
- Google is UI: Stack Overflow was built around the assumption that people will go to Google which will send them to the right page. Each URL has the name of the question; each URL is permanent and clean, Metatags, sitemaps; anything and everything was done to ensure Stack Overflow's pages looked "reasonable to search engines."
- Performance: Ensuring answers are provided super fast was imperative. As a result, Stack Overflow is built on a Microsoft stack. "This entire site is serving 16 million pages a month and we're doing it off of two servers which are almost completely unloaded," said Spolsky. One server is a Web server, the other is running Microsoft SQL Server 2008 and both are 8 core Xeon's. While many may assume using an open source stack would be more efficient, Spolsky explained that while SQL Server licenses cost $5000 per box, the Microsoft stack is paying for itself in terms of reduced hardware.
- Critical Mass: It's imperative to have critical mass on day one; to ensure people are available to answer questions. "That was one of the reasons I asked Jeff Attwood to be involved in the site," Spolsky explained. Between Joel on Software (Spolsky's blog) and Coding Horror (Attwood's blog), the two had a combined visitor count of 1.3 million visitors per month. Combined with the weekly podcast the two began, they were certain to get at least 20-30 thousand programmers interested.
We've embedded Spolsky's talk below and it is well worth an hour of your time; particularly if you're interested in building, or have already created a social site; Stack Overflow's numbers speak for themselves.
Share
Facebook
Comments
Subscribe to comments for this post OR Subscribe to comments for all ReadWriteWeb posts
Thank you - very insightful, permanently bookmark and will discuss at my digital marketing academy.
This is an insightful look into the workings of an intriguing and useful social site, and there are certainly plenty of lessons to be learned here. These are all reasons why Stack Overflow is useful and well designed to accomplish its core objective, and they could be reasons why the community sticks around and the site continues to thrive (if it indeed does that)... but none of them are the true reason Stack Overflow got so big, so fast. The answer there is most simply: celebrity.
Both Joel Spolsky (Joel on Software) and Jeff Atwood (Coding Horror) had considerable celebrity among their target demographics. Compete estimates about 158,000 uniques per month at eith site (my guess is they get more than that) and both link to and often talk about Stack Overflow.
People paid a lot of attention to Stack Overflow from day one because of who was behind it. It definitely couldn't suck (and it doesn't -- for the reasons you mention in this article)... But the answer to your "How?" question that's closest to the truth probably has more to do with the celebrity of the founders than anything else.
Amazing stuff and worth reading twice - the power of new media platforms (social media channels) is that you empower users and create a community feeling to awesome results.
"Ensuring answers are provided super fast was imperative. As a result, Stack Overflow is built on a Microsoft stack."
This claim sounds highly bogus and sticks out like a sore thumb among the otherwise solid advice. The 'millions of pages' per month quoted come out at only 6 pages per second. I think we could all manage that given two 8-core servers loaded with memory, even without the mystical benefits of an MS stack.
I believe SO also uses memcached, which last I heard was an open source component. That and the fact SO's servers are stuffed with memory and CPUs probably has a lot more to do with their success in serving pages quickly than has SQL server.
Seriously, is this product placement?
Yep, the real choice was probably a bit like this : Jeff know ASP best... and that's all.
Interesting post.. what a very challenging video. Thanks
This entire article could be cut down to "find a target audience [Joels community], solve a problem [I can't ask coding questions on the web], reach said target audience [via Joel and Jeff]".
While all these points are the reason the sites lasts, the reason for success (as comment #2 points out) is point #9: Critical Mass.
You could be the smartest programmer and the best anthropologist in the world, but you can't think up Critical Mass.
Takeaway: If you want to start a community, make sure you do it in a field where you are a respected expert with a following (or partner with a respected expert). If not, start a blog, build a following first, then launch your community.
First off, Joel means ethnography, not anthropology. Anthropology is the study of societies and cultures, whereas ethnography is the study of "the customs of individual peoples and cultures". Specifically all of us interaction designers use that to talk about the way people interact with their environment, in reality.
Sadly, the rest of the premise is a joke. Does the site use social mechanisms to shape behavior? Sure. Does that help shape the quality of content? Probably.
But is that the secret behind its (theoretical) 1/3 audience coverage? Could another person duplicate that? No. It's a function of fanboys, and the audience they chose (nerds).
That two super-famous people who have legions of fanboys in a field which is super impressionable and addicted to shiny new things, could create a social site which attracts a lot of people... not a surprise. But it can't be broken down to individual points of success which another person could use to duplicate that success.
Yeah, curious bit of plugging for MSFT, but several other points well scored. (Will be interesting to see Calacanis' thoughts as well as the gathering throng of Open advocates...)
My fave is 'Google as the [initiating] UI', increasingly a universal truth for almost all websites that are not yet discovered, community or otherwise.
But the other 8 building blocks could easily have been a different '8'... glaringly missing are things like 'Community/Stickiness', 'Accuracy/Relevance', 'Sharability/Openness', 'Authenticity/Authority'.
Perhaps RRW holds a little social media experiment and asks readers and social engineers alike to submit (vote, tag, ; > etc.) their nine most critical buildings blocks for social media site success.
I'll translate this article into russian. It's very useful.This is my opinion.
Disappointing article from RWW in its celebrity worship.
And what about http://ask.metafilter.com, which was the first of the answers sites and another in which the quality has consistently been high?
great post and the video is on my to watch list this weekend.
For those that are calling BS on the MS stack I will ask how many of you actually use it? I have always found .net/iis/mssql can take a huge volume of traffic out the box.
Actually, Stack Overflow DOES require registration.
And he's right, it does reduce participation. I was going to post, but close the window when I realized I'd have to register.
Ethnography is part of and study of anthropology. What this guy is describing is more a social constructivism, a man made sociological, not anthropological, phenomena.
Very insightful and a good list of options to make your site more interactive!
I'll be reading this more than twice, I think!
An anthropological approach to social media would be to make a case study of a particular site or blog, observe a lot, participate a bit, and write about the social interactions seen. The emergent hierarchies of the most popular or talented; social regulation through overt rules (e.g. acceptable topic stipulations) and social norms; disruptions, such as those by the insufficiently socialised (trolls). What functions does the site provide for its users? What complexities elsewhere in the world/their lives are mediated/mitigated for users by their participation? How does the social economy of the site interact with other social factors (class, economics, gender, race)? What, besides forum/blog comments, does this participation and social space produce - especially, what meanings? This would be the basic anthropological approach to a social website.
Reverse engineering social interaction - as done by Stack Overflow - isn't anthropology, as Amy and Cooper have said. Two reasons: wrong way round (anthropologists watch something already in existence, rather than invent a social scenario) and because it's not a cultural but a behavioural approach. You might be able to call it applied social psychology... But that wouldn't be such a catchy headline.
Wow, some of the commenters here sound really threatened that 2 mere programmers would come up with a combination of social factors that made for a successful site. As a user of Stack Overflow from pretty early on (I guess that makes me a "fanboi"?), I can say that -- with the exception of the ridiculous claim that only M$ technologies could possibly power this site (2 8 core servers, of course, being the equivalent of SIXTEEN servers from a few years ago!) -- this article pretty well sums up what has made SO successful to me.
The buzzwords tossed out by @Thom are too vague to be useful. The snarky dismissiveness of @Amy just smells of jealousy. And, @Daniel, critical mass isn't sufficient to make SO useful. Yahoo! Answers, AskVille, Fluther, etc. have critical mass. It's the quality of the answers that make SO valuable to programmers.
In programming, the knowledge is so arcane and specific, if you're struggling with a problem, you may need to find the 2 or 3 people *in the world* who have run into the same problem before *and* successfully solved it. It's quite possible that your problem has never been dealt with, but you need a community of people with similar skills to help you find a creative solution.
That's not going to happen on those other sites.
Anyway, I find it amusing that non-programmers are bagging on Spolsky's depiction of SO's special sauce. 8 of the 9 "building blocks" (and, actually, performance does, too, just not that it has to be powered by Microsloth) have face validity to me and, importantly, are specific enough that someone wanting to build the Stack Overflow for, I don't know, sewing patterns could take them and implement them on their site. Badges? Yeah, I can implement that. Stickiness? Um, not so much.
For completeness, here's what I had to say about Stack Overflow immediately after asking my first question on the site: http://andrew.hedges.name/blog/2008/09/20/stack-overflow
I actively use Stack Overflow, and I'm a big fan of the service they provide. I also agree they they nailed the social parameters for their community really well.
However, I agree with most of the criticisms in the comments here. The biggest factor is the influence and popularity of the founders in drawing attention to the site. Having said that, I actually found it first through google. Their SEO is very good, and will only get stronger with time.
The claims about MS architecture vs. Linux/MySql, or whatever else the comparison may have implied, is completely off base. To be credible, publish real benchmarks comparing apples to apples. For that kind of stuff, unprovable anecdotes are not necessary. It takes away from the real point: Stack Overflow is a great site for programmers.
Cheers!
Walt
@Josh Catone
Exactly! Their celebrity has to do a lot with the success of SO...
I'm wondering if those two would be nobody programmers (good professionals but still nobody) SO would have had the same kind of success...my guess is not.
Andrew: Threatened? I don't see that. The criticisms I see here are not so much of Joel and Jeff as they are of the article, which takes a list of pretty basic good ideas that everyone knows about, wraps it up in an impressive-sounding term ("anthropology"), adds a bizarre non-sequitur about Microsoft, and treats it as a publishable insight. The fact that some folks find the article somewhat underwhelming doesn't (a) diminish its usefulness to people who hadn't thought about these issues before, and (b) doesn't constitute disrespect to its subjects.
Nice link-whoring though.
I noticed the lack of ask.metafilter myself. Perhaps part of the reason the quality has always been consistently high is because they don't play games with their users to try to get traffic. There's no rating or badges or any of that crap there. There's a "mark as best answer" and a few flaggng options and that's about it, all of which are very a minimal part of the interface.
@Name, um, I think you misunderstood. SO doesn't require registration to ask or answer questions - it does require some rep to do things like upvoting or commenting, but for the two primary functions you can be unregistered.
This is, in fact, seen when some users are identified as "anonymous (google)" or "anonymous (yahoo)". These are people who keep returning to the site but haven't registered. The site is totally usable if you aren't registered.
@Rod, Good points, except for the link whoring comment. That's just lame. Makes perfect sense to me to give more context to my comment by linking to a blog post that describes more fully why I find SO valuable.
Nice name calling, though.
StackOverflow is useful.
However, it has as much to do with anthropology as it does with Chinese take-out. Maybe less.
These poor people need a basic liberal arts education. I suggest all StackOverflow users leave tips to finance this.
I think this is the perfect example of Seth Godin's Purple Cow. Challenge status quo, do something different and you will win.
I'm not the one who started throwing "threatened" and "jealous" and "snarky" around, or linked to a four-paragraph blog post that says nothing, really, that hadn't already been said here.
Anyway.
It seems to me that an ignored factor that makes the answers on SO better, is that that the questions are better. The subject matter itself creates a barrier to entry that weeds out most of the Yahoo Answers-level questioners (and answerers). Ask better questions, get better answers--to me that explains more than "badges" does.
We actually took a page from Stack Overflow's book about a month ago when we decided that downvotes should cost each user 1 reward point at CreateDebate.com.
Anthropology covers a wide field and cannot be boxed into this or that. It also covers Physcial anthropology where individuals have to learn about the skeletal framework to understand, the physical activities, posture, etc of individuals, when they carry out diff jobs. So let's not bicker here about what the word anthropology here in this article covers. It could also have been used metaphorically speaking, to cover human actions, not necessary the user or the initiator.
Secondly, yes "celebrities" do have their role to play in "branding" and getting the "sales up". But if its a good thing, it does catch on. If the public does not like it, no matter who dishes it out, it just doesn't catch.
Thirdly, the "modus operandi" used here is I think the most important to understand, be critical appreciative, and not just write it off and brush it under a carpet.
The kind of voting mechanism, and that besides the community, even the one who has submitted the problem and which of the solutions was the best fit, has the say, is what I think has "anthropological underlying meanings" (be it ethonography, cultural anthro, behavioural anthro:lets stop splitting hairs here). An example of this would be village situation wherein a problem is taken to the community headman, in front of all the community. The headman does not think of it himself, but with the community involvement. Many solutions are forwarded, the best fit is accepted by the "initiator". The village recognizes that but also accepts the rest of the solutions as possible ones that others that could be used to solve similar other issues of their own, and everyone has gained here and learnt here. Evolution at its best ever visualized in a small "tool" that has helped win over the other tools that have yet to capture your imagination.
Anthropology covers a wide field and cannot be boxed into this or that. It also covers Physcial anthropology where individuals have to learn about the skeletal framework to understand, the physical activities, posture, etc of individuals, when they carry out diff jobs. So let's not bicker here about what the word anthropology here in this article covers. It could also have been used metaphorically speaking, to cover human actions, not necessary the user or the initiator.
Secondly, yes "celebrities" do have their role to play in "branding" and getting the "sales up". But if its a good thing, it does catch on. If the public does not like it, no matter who dishes it out, it just doesn't catch.
Thirdly, the "modus operandi" used here is I think the most important to understand, be critical appreciative, and not just write it off and brush it under a carpet.
The kind of voting mechanism, and that besides the community, even the one who has submitted the problem and which of the solutions was the best fit, has the say, is what I think has "anthropological underlying meanings" (be it ethonography, cultural anthro, behavioural anthro:lets stop splitting hairs here). An example of this would be village situation wherein a problem is taken to the community headman, in front of all the community. The headman does not think of it himself, but with the community involvement. Many solutions are forwarded, the best fit is accepted by the "initiator". The village recognizes that but also accepts the rest of the solutions as possible ones that others that could be used to solve similar other issues of their own, and everyone has gained here and learnt here. Evolution at its best ever visualized in a small "tool" that has helped win over the other tools that have yet to capture your imagination.
Herr Atwood's name is spelled with only one 't', not two.
The whole article was basically taken from Joel's mouth. You might was well call it an interview, to avoid the impression that there's any independent thought or analysis going on here.
very useful! It's quite possible that your problem has never been dealt with, http://www.vcao.net but you need a community of people with similar skills to help you find a creative solution.
great article,thanks for you to tell us this information~
"What we do have to think about [in the era of social networking] is human to human interaction," he said. And according to Spolsky, to do that, you have to think as an anthropologist does.
Whoa you sure have Much comments!
We must accept finite disappointment, but we must never lose infinite hope.
it can't be broken down to individual points of success which another person could use to duplicate that success.
I'm wondering if those two would be nobody programmers (good professionals but still nobody) SO would have had the same kind of success...my guess is not.
nice site with lot of informative information.thank you very much for sharing nice information.
Very useful information, as for me. How can i download this video frim youtube service? Thanks.
I will make sure and bookmark this page, I will come back to follow you more.
Thanks a subject that truly move
Hi,
Good work. Thanks for sharing.
Hi,
Good work. Thanks for sharing.
iam very glad
Supposed to attack these head-on and you will find a deep sense of gratification thatwill fuel your happiness.
It seems to me that an ignored factor that makes the answers on SO better, is that that the questions are better. The subject matter itself creates a barrier to entry that weeds out most of the Yahoo Answers-level questioners (and answerers).
It is necessary that we should absorb some essential results of Anthropology and the Social ,to make our future more brightly
This is great! Really. I'll make my own little contribution to the list soon
This is great! Really. I'll make my own little contribution to the list soon
1 2 Next