Raising money for StackOverflow(joelonsoftware.com) |
Raising money for StackOverflow(joelonsoftware.com) |
On the one hand, it's definitely better than ExpertsExchange, and does an adequate job of providing answers to many developer questions.
On the other hand, it's far (far!) from living up to it's potential. The basic premise of the site is that the wisdom of the crowd will separate the gold from the dross; however, the incentives lead to a plethora of mediocre answers which do not converge (Wikipedia-style) towards some platonic ideal, despite Joel's stated wishes to that effect. What's more, there's a well-documented problem ("the fastest gun in the West") whereby the first semi-correct answer will quickly rise to the top, drowning out later (but more correct, more thorough) answers.
In short: they haven't (yet) solved the basic problem of "changing the way people get answers".
What's more, gaining the critical mass of early users was leveraged largely through Joel and Jeff's celebrity. Which is not transferrable to other domains. Which means that the site can't easily scale out.
Based on the preceding, I'd say that VC money is a good way to go-- it should give them the resources to a) solve the basic problem, and b) build subscriber bases in other domains.
More and more when I search for solutions to problems I'm having I've been coming up with SO links. And when I do and the SO question is related to my issue I usually get the solution, too.
A mediocre answer that was posted one minute after the question was submitted often has more votes than a well thought answer posted several hours later.
A better solution might be to randomly order answers, regardless of votes, for the first couple of hours (maybe even day). This will further help remove any ordering bias.
I also feel that authors of answers should be hidden for the first couple of hours (or day). Just because John Resig answered a question, doesn't necessarily mean it's the best response.
The worst case scenario is a question which is perfectly written such that it garners attention from experienced folks who know what they're talking about for only a tiny amount of time, and ever after only garners attention (via google searches, for example) from clueless noobs who can't judge the quality of the answers (and don't have the rep. on the site to vote answers up or down anyway). However, this is a pretty contrived scenario which doesn't play out often in real life. Important questions will garner attention by people with a range of experience and qualifications, which will have the side effect of improving the quality of answers over time through future voting etc.
Of course, the main problem is that individuals should be able to license their contributions as they choose.
Also, it doesn't matter for most code sample on there. Most are not anything you would want to copy (too specific, or too general, to apply directly to your problem), or short enough that copying them would be considered fair use.
I haven't noticed this. I know I have personally answered questions months after an answer was accepted, and have gotten plenty of points. In some instances, the "accepted answer" was even changed to mine.
Stack Overflow's problem is that 80% of the answers are from talkative people with no clue, and 100% of the questions are from people who have never heard of Google.
The more germane question is, are the problems best fixed via an infusion of VC cash?
In this case, I think the answer is "yes".
If I come to Stack Overflow with a question, many times a quick, off the cuff but not perfect answer will solve my problem. Sure, it might not be the best possible answer, but if it solves my problem, why would I want to look for something else? The primary person getting value out of an answer should be the person who asked the question, or else no one will bother coming to ask their questions. If you put anything in to slow down the submission of answers, you reduce the value of the site to people asking questions.
And there's an easy solution if the top rated and accepted answer is actually incorrect or misleadingly incomplete; just edit it. Sure, it takes a bit more rep to be able to edit people's answers, but it doesn't take all that long to get that rep.
incenting 'cleanup' (the original intent of community wiki) would be great, but i doubt it would increase their traffic/revenue very much.
I get the impression Joel's looking at expanding the SO model into other non-developer/non-geek niches where advertising could work, but I'm not convinced he's proven the revenue model enough to make it a tempting investment just yet. Anyone know any different?
Note: On another post, someone reminded me of the Careers move, so that's one easier-to-monetize stream at least.
They don't need to care how I monetize my sites.
[Sadly now they do not allow this in any way]
To somebody like me who doesn't really understand advertising, it seems like the value of an impression ought to be really high for StackOverflow. Based on the tags associated with my userId and the questions I've read, they pretty much know which technologies I use everyday. How much would it be worth for EnterpriseDB (the PostgreSQL folks)to advertise to someone who is wondering how to covert his Oracle database to MySQL? Perhaps, even with 6M users, there are so few people in any coveted niche that the friction overshadows the value?
Advertising on something like SO is almost like advertising your latest application on the Pirate Bay. I just don't see the value.
Jeff Atwood seems to get a lot of flack, but in creating such a nice site and community, he's walked the walk as far as I'm concerned.
If it's marketing, I don't see why they can't do that with existing revenue. It's not as expensive as say a retail location for a starbucks and they must be doing some revenue with ads and job board already.
If it's hiring more developers that might make sense, although I can't think of any huge features they don't have yet.
Overall, not sure I'm convinced this is a good play for VC. They are growing nicely without it.
StackOverflow has launched what, 5 QA sites in total now? All of which seem to be thriving pretty well. It seems they're expecting similar success with QA sites for Sailing, Pregnancy Advice, Cat Owners, Subaru Mechanics, etc.
Have they? Whenever I have a question the first place I turn is Google, and I usually find my answer on a site that's not stack overflow.
Craigslist is widely described as leaving 90% of the money on the table. Most people don't think this is why they are this popular. They could have monetised more categories or cities and still kept their position. Maybe.
But maybe they did exactly what was necessary to become the number 1 classified site. What if the right amount of revenue for some top 200 site (or network of sites) is really a couple of million?
Can such a site survive VC or does the option of staying a certain size disappear??
Starbucks was a perfect example. They new exactly where the business was going if constraints got lifted. More stores. Potentially tens of thousands. If they were less capitalised they would have to go slower but they knew where they were going.
How it goes: I post a question(questions are rather specific, but technology stack is very common on SO)
I usually get one-two answers very quickly and maybe one more slightly delayed answer, all helpful but also rather generic. You definitely get the impression people are just point whoring.
After a day or two I end up accepting the closest answer, even if it is not that close.
Seems real experts have moved on or perhaps I am missing some trick in asking questions.
If this problem persists on their scaled out sites, I am not sure how much VC money will help them.
Of course VC makes sense in their case. There is no barrier to entry in this field, the model is clear:
* users tend to type questions in search engines
* search engines tend to rank up exact matches (especially with some trivial SEO like put the question in the title and url)
* given the proper environment, users will help each other answering those questions for free, generating lots of content - exposing ever bigger surface to the search engines
The question I'd be interested in knowing is, "if SO was in accepted into a YC intake, what advice would SO be given to evolve given the app is built, idea is validated?".
Presumably if we found some interested parties, they would want to meet the key team members, in a second or third meeting.
He doesn't talk so much about what a successful badass he is (see Joel), despite obviously having accomplished some pretty major shit.
Clearly, based on their traffic, they have changed the way many, many programmers get their answers.
Look at how StackOverflow has displaced Experts Exchange, though, and it tells a clearer story:
http://www.alexa.com/siteinfo/stackoverflow.com+experts-exch...
I don't think there is ANY other general-audience programming site with more traffic.
http://stackoverflow.com/users/5020/melling
I'm going to turn Emacs Lisp into a mainstream developer tool. :-)
Actually, the best technique I've found is Google with "site:stackoverflow.com", as Google does a much better job searching StackOverflow than the built in search engine.
Either you get a successful site in that vertical, or somebody else pays you $50k/year for the right to have it instead.
If a site gets to the point where it is a substantial business, they may well want to partner with Stackoverflow more closely anyway to get features, advice, access to whatever revenue generating plan they come up with.
StackOverflow should seriously consider using this algorithm, at least for a day since the question was asked, as one previous commenter said.
Optionally, question submitters could pay for "Urgent" status, where the submitted answers are posted as they come in. No ratings or votes would be accepted for the same cooling-off period, however.
For example, I tried to get the Grails community to use StackOverFlow when the site first started and their attitude was that they would rather just use the mailing list and sift through Nabble.
http://archive.codehaus.org/lists/org.codehaus.grails.user/m...
The problem isn't with StackOverFlow, it's with the other communities refusing to adopt something new. "After all, IRC is where all the best people hang out...who needs this .Net thing..."
I'm currently learning Lisp. You wouldn't be able to get this kind of an answer on a mailing list or on IRC.
http://stackoverflow.com/questions/2264267/generating-a-quiz...
This .Net/C# site just made it easier to learn Haskell, Lisp, Erlang, Scala... a little irony not to be overlooked.
The first sentence is fine. The second not so much. "Fault" assumes that moving to SO/contributing etc is unambigously a good thing. I would not be suprised if at least some developers thought their existing communication channels are just fine.
"For example, I tried to get the Grails community to use StackOverFlow when the site first started and their attitude was that they would rather just use the mailing list and sift through Nabble.
http://archive.codehaus.org/lists/org.codehaus.grails.user/m.... "
This sounds like a valid choice to me.
"The problem isn't with StackOverFlow, it's with the other communities refusing to adopt something new."
They may have good reasons to do so. Just because you are enthusiastic about SO doesn't mean everyone has to be.
(Due Disclosure : I don't use SO at all. I once saw an algorithm question on SO I could answer and when I tried, I found out I had to use OpenID to log in. Couldn't be bothered.)
The Perl community has a decade of history and over 800,000 posts of history invested in http://www.perlmonks.org/. Granted the site is creaking and could use a better design, but someone with a Perl question is likely still better off going there than to StackOverFlow.
If someone ran a malware-ad campaign on SO, I'm sure a significant number of people would be infected.
I'm actually surprised by how many programmers DO have adblock.
Check out the FEBE and OPIE plugins for firefox. Host the archives someplace (dropbox works well) and write a little install script and you're done.
Installing stuff by hand is so last century...
People tag their questions, while others vote on the questions and answers. Question titles are matched against the database when people enter new questions. Admins "close" duplicates, and stop trolling. The reputation "economy" provides a way to get an answer.
StackOverflow isn't perfect but it is a big step forward in organizing the information so that that the data can be searched more effectively.
http://stackoverflow.com/search
There are 500,000+ questions tagged. Some are great, some are not. Joel and Jeff don't provide either of these. We do!
Don't try this on a mailing list.
http://stackoverflow.com/questions/101268/hidden-features-of...
On Ruby Inside, Rails Inside, and Ruby Flow, I run a cpl million ad impressions each month and make an OK amount of money from it despite the CPM being deathly low (say $2-$3 on a per ad basis). The CTR, however? It's south of 0.5%.
I also ran an Adsense for Feeds experiment for a few days recently. Don't have the numbers to hand but it was 1 click in approximately 16,000 impressions.. and these are real, unblocked impressions. I made 40 cents. Developers don't click on ads. "Geeks" do, though, IMHO.. otherwise sites like Slashdot wouldn't bother.
Compare this to Adsense ads on my old, archived personal blog.. and they run at over 20%.
The less web savvy the audience the higher the click through, best CTRs I ever experienced were on myspace related sites, my target audience was 15 year old kids judging from profile links I saw.
As for Slashdot, I always sort of wondered whether or not the appeal to those advertisers was less clickthrough and more name recognition. That is something that I think is a bit under-rated. I think that matters when evaluating products - I know that I tend to favor brands that I've heard of in some way vs something completely unfamiliar.
That has been cited by some advertisers, for sure - especially those who are recently funded or have larger branding budgets.
Take New Relic, for example. They're sponsoring The Ruby Show for I'd guess (and I don't know for sure, but basing on rates of similar podcasts) at least $300-$500 per show? There's no way they're getting that much business from people listening to the show checking them out at that very moment. In terms of brand awareness, though, $5-8k a year is nothing if almost every Ruby developer has at least heard of you..
I think it's rather "geeks don't see ads".
On a related topic from that link, I don't think I've ever come across a banner ad made to look like a dialog box that has been good enough to fool me. But again my experience seems to suggest that those with less computer experience are far more easily fooled.