I launched KDE-Look.org, the first website in the openDesktop.org network nearly 15 years ago. This is a long time and I announced today that I sold the network to Blue Systems. So I think it is time for a look back at how everything started and where we are today.
How it started
I became really interested in Desktop Linux around 1996 when a friend of mine showed me a beta of KDE 1.0 running on SUSE Linux. I was blown away by the fact that is looked as good and powerful as Windows 95. And that it was developed as free software with an open process by a community and not by a company. I started to read the mailinglist, scriped a small build script that compiled the latest repository checkout every night and got slowly more and more involved. On aspect that fascinated me was that it was so flexible. You can use a ton of different free software applications that could all be customized and tweaked in different ways. A big part of that was the themes and artwork that could be used. The leading website to get these themes and tweaks and small patches was themes.org. The problem was that it was constantly down and honestly not very good.
I was running a web development team at an internet agency at the time so I had some experience on how to to build something better. I came up with the idea to launch a new website to fix the problems and also stimulate and grow the open development process more. I still remember how I scribbled the first screens and architecture during a vacation in Greece. After some coding in my spare time I launched KDE-Look.org in summer of 2001. I had the opportunity to host a personal web project at the servers of my employer for free because I was also responsible for a small hosting center for our customers.
Crazy growth
The problem was that KDE-Look.org was immediately super successful. In just a few days the website generated more traffic and page impressions then all the other websites in this hosting center combined. So I had to look for a new hosting option. A provider in Stuttgart offered me free hosting if I brought my own server. So I bought a 19″ server and ran it in that hosting center for a few months. The problem was that the traffic kept on growing so that they canceled the deal and I had to look for another option.
For years I was constantly looking for affordable best ways to run this. Remember that this was still only a hobby project that generated multiple TBs of traffic per month. I had to pay everything from my own pocket. I approached the KDE e.V. but they didn’t want to host user generated content for legal reasons.
The problem wasn’t the pure data transfer traffic alone. The website also generated millions of page views per month. So I invested a ton of time in fine tuning the PHP application including the database queries and a sophisticated caching system. At the end the websites requires less then 0.5 SQL queries per page view because of a very sophisticated (and complicated) caching system I wrote. All of this on a completely user generated and personalized website.
At this point in time I decided to add advertising to the website because I couldn’t afford the hosting costs anymore. This generated some money that covered parts of the cost. But more about this later.
More sites and collaboration
I have always believed that Free Software projects should collaborate more and work more closer together. I never liked the fight that was going on between KDE and GNOME and between other projects in the open source community. GNOME was also in need of an artwork community and when someone asked me if I could help I decided to launch GNOME-Look.org too. But I did it in a way that the user accounts are shared and even some content categories are shared between the websites. For example I shared the wallpaper categories because they are clearly not desktop environment specific. Turns out, I was wrong. A lot of people freaked out when they say a wallpaper with the words KDE on it on GNOME-Look.org and the same for GNOME branding on KDE-Look.org. So unfortunately I had to remove most of the content sharing between the websites.
In 2003 the website that hosted 3rd party applications for KDE went down. So I launched a replacement under the name KDE-Apps.org which was also very successful. Over the time more and more communities and projects approached me and I launched websites for them. All on the same platform with the same multi tenant system. At the moment there are around 30 websites for different projects.
The traffic and the page views exploded and it became more and more expensive to host the cluster of several webservers with TBs of traffic per month.
Web 2.0 and a new company
Then in 2005 – 2006 the term Web 2.0 came up. This was the time when the startup world recovered from the 2001 .com bubble crash. A lot of companies where founded to build new and innvoative web businesses. Facebook, Twitter and lot’s of other companies became big. I looked at some of this new companies and noticed that my websites generated significantly more page views then some of these “cool” new communities. At the same time I was still doing all my open source activities as a hobby in parallel to my day job. So I came up with the idea and dream that it would be so cool to be able to do all this work full time. The websites generated some advertising revenue. Not enough pay for the servers and for me but I was doing this mainly during the weekends. So I figured that I could do a lot better if I would do this full time. So I decided to jump in at the deep end and give it a shot. I had luck: during the meeting where I told my boss that I quit to start a company he offered me to invest in it. So I founded hive01 gmbh, my own startup in 2007.
I started openDesktop.org as head website on top of the 30 others and managed to grow openDesktop.org faster then before. I hired a few employees and rented an office in Stuttgart. Unfortunately growing the advertising revenue was not so easy. The reason was the financial crisis and the unwillingness of companies to book advertising on open source websites. Another problem with the growing number of users with ad blockers in the open source space. A discussion that is very hot today again in the mobile space. We licensed the software to IBM, Nokia and a few others which was great. We managed to get close to profitability but never really reached it so I was slowly burning through my investment and private money.
Crazy new ideas
Over the years I came up with all kind of different ideas. Some of them where good, some of them not so much but it was a lot of fun.
For example my websites where a real social network way before the term social network became mainstream with friends, messages, activity streams, groups and user generated content. This was all before Facebook or Twitter.
Then I launched the first App Store where developer could sell free software to users way before Apple, Google or anyone else did this.
Then I launched an innovative job board where developers and companies could be matched way before this idea was cool.
A user generated event database, a knowledge database before stackoverflow and other became mainstream, a user generated open source news portal for blog posts, a short messaging system similar to Twitter and a ton of more stuff.
One interesting idea was project Bretzn where we implemented a way where a developer has the option to publish an app from the IDE QtCreator directly to the openSUSE Build Service for cross compilation and to publish it on an AppStore to sell it with just one click.
Another project we did was the Open-PC project, which had as goal to build a real open and userfriendly Linux PC as a community project.
Another initiative was to build a decentralized way where Desktop and Mobile Operating Systems like KDE could integrate social features directly into rich client Desktop applications. Very similar how most mobile application are frontend on top of web-services nowadays.
All these initiatives where a lot of fun and they all where more or less successful. But in the end it was always tough to generate the necessary revenue to keep the company alive and pay the employees and interns.
Coming to today
While releasing a lot of free software code and contributing to free software in a lot of different ways I never released the openDesktop.org source code under a free license. The reason is that while I deeply care about free software it was my number one priority to keep the websites and later the company with its employees alive so I could keep running interesting, fun projects which would bring new services to the Free and Open Source ecosystem.
I’m proud that we never had a data loss or a hack or serious security problems or a longer downtime. Especially if you consider the size of the websites and the amount of traffic they generate. I’m also happy that we where able to release and run very innovative and useful features for the open source communities to stimulate the open collaboration here.
Today, openDesktop.org is in a good spot, though I can think of 100 things I’d like to improve including the frontend design. Unfortunately I haven’t had the necessary time the last few years to push the network forward because ownCloud consumed all of my time.
But I think that I found a good solution for the future of openDesktop.org as mentioned earlier today.
Thanks a lot for reading this long blog post and let me know if there are any questions!