SWiK needs a reboot
It’s been over a year and a half now since I quit working on swik.net.
Working on SWiK was a weird experience. It was a total experiment, can we build a community of people to help other people get into open source software, or figure out how to actually use open source software.
Fortunately or unfortunately, SWiK started to pull in Google referrals by the truckload. This did not translate into dollars however, while I worked at SourceLabs we did not have advertisements on the site.
Instead of advertising, I tried to develop ways to incent all the people coming in from Google to help develop the content of the site, to write about open source software and develop the wiki. The concept of the site was to let users do the content generation and me and another developer would work on the wiki software, which we wrote from scratch.
We tried a lot of things, but people coming in wouldn’t convert to community members. We tried zero registration, getting press about the project, we tried letting people have personal pages, personal tagging, we tried ‘promote your project’, we tried building sample high quality pages, we tried changing the site copy, lots of things but none of these worked. It was very frustrating, but oddly so since every day we had more and more people coming to the site.
A lot of our time was actually dealing with all the Google traffic. We quickly got up to 2 million hits a day, on a system that was designed for an order of magnitude less volume. Another big time sink was dealing with incoming spam. Having an open wiki with no registration meant that 99% of “contributions” were spam, and we tried to develop tools and tricks to eliminate them without annoying the rare good contributors to the site.
SWiK still generates a lot of traffic, over a million unique visits a month but still has no community. These visits are all from Google and to tell the truth it’s annoying to get swik results. SWiK has some good pages, but Google sends the majority of traffic to the crappiest pages. We would be on the front page of Google results for ‘www.bebo.com’ because some stupid person came in and edited www.bebo.com. Or Google would send a million people to a million different random links we had in our tagged content aggregator.
So what did I do wrong? I’m still not sure what would have been the right design, the only way to know is to put something out there and for it to work, and even then there’s probably a degree of luck to getting a community to coalesce around a project. We had competitors trying the same thing as us and they all failed too, so I don’t feel like the solution is completely obvious.
I still want the site to work, to provide a good resource for open source information. To work, I think the site needs to start over and keep it simple, stupid. Delete all content. Rewrite the engine. Focus this time on something concretely good instead of theoretically good: high quality articles on open source software. Make more barriers to publishing, and try paying for or otherwise rewarding high quality content. Focus the feature set down by cutting almost every feature: all the site should do is the most simple thing: high quality pages about open source software subjects.
This won’t happen of course. For the same reasons I didn’t do it when I worked on it:
- No money to pay for content.
- No money to pay developers to rewrite the engine.
- Don’t want to risk an asset.
Recently I’ve built a site engine that can do the same thing as SWiK in terms of SEO, and I’m thinking how do I avoid the same mistake, don’t I have a responsibility to not push SEO too far?
This time around, I will try and limit how much I do and learn from swik, but as I intend to also sell the engine to people who want to use it to promote their stuff, it will be a bit tough to control how it is used.
