Eerlijke informatie over Linux
It is very customary to run a game show at the end of an ELC conference. For the Americas, the show is always hosted by Tim Bird. I decided that I wanted to try to do it for the European ELC conference. So, for the second ELC conference, held November 6+7, 2008 in Ede, the Netherlands, I planned to run "Family Feud" - which has been airing around the world in one way or another. Note that Tim already had done "Feud" on a CELF conference before.
The original formula of the game involves two teams of five. In preparation for the game show, the producers have asked the "population" a series of open questions and categorized the answers. It is the goal of the game that the teams guess which answers occurred most.
In the original game, the end game ("Fast Money Round") either is successful - earning the team a money prize, or unsuccessful - in which case the team goes away with nothing. As we were only running this game once, that was no option. I also had only one prize - so we needed to modify the game to come up with a single winner. We started out with 4 against 4. The winners split up and ran another series of questions 2 against 2. The winners played against each other in a variation on the fast money round.Looking back, I should probably have stuck to 4 questions in the first round and 2 questions in the second round. That would have reduced the time spent on the game - it now ran for nearly 80 minutes.
For his version at the ELC 2003 conference, Tim had written some client/server based scripts. According to him, it didn't work really well. So I decided to search the internet. Apparently the owners of the game formula are quite aggressive in taking down home-brew score boards - I had to search quite thoroughly to find one. Note that i ran into many suggestions to just use powerpoint. In hindsight, that might have been a better idea...
During the rehearsals, it all worked fine. However, during the real game, it failed miserably. I've learned a few things from this:
According to the audience, it didn't bother them much that everything failed - most just enjoyed me mess up.
I started out in a spreadsheet, but I quickly discovered that spreadsheets don't work. So I wrote an application in QT, which I already know from my open source project for enabling 3G UMTS devices in Linux, umtsmon.
Designing the data model took me an hour, implementing it another four. Adding the various tabs in the GUI was done at a later stage. Note that several classes were directly copied from the umtsmon project, saving me implementation time. In total I think I spend 8 hours on the application. The sources are available here - make sure you have the qt3-devel package (or similar) installed and use the QT3 version of qmake to create a Makefile or you'll have big problems. I started out with 20 questions, but soon realized that some of the questions really didn't work. The results from the 18 other questions are below. In total I interviewed 49 persons from 13 countries.
This was a simple question, just to set the mood. I was amazed at the wide variety of answers and the little amount of people who came up with "any".
The second question was meant as a deeper dive - to show people that I really just wasn't going to just do "funny questions". Several of the interviewees couldn't name a single directory or didn't understand - other than that, I think the answers are good and to be expected. I marked the wrong answers with an *. If you want to check, there are various LXR sites around the world that let you easily compare/crossreference the Linux kernel source.
These answers actually surprised me - I hadn't expected ls to be so widely dominating. That's why I decided to skip this question - the results did not fit the statistical profile needed. Fortunately many people were sharp enough to just say busybox - which is obviously correct.
It actually was quite fun to ask this question. Most interviewees would blush, some downright refused to answer the question. I hadn't expected this question to be so loaded. Anyway, the results were fun - but there are many answers I suggest one shouldn't try in real life.
|1||can I buy you a drink?||6|
|2||You look beautiful!||5|
|3||Have I seen you before?||3|
|4||what distro do you run?||2|
|5||how are you doing?||2|
|6||Would you like to pet my <pet>?||2|
|7||Hi, nice shoes!||2|
|8||your hair looks good!||1|
|9||what was your last hack?||1|
|12||do you come here often||1|
|13||did it hurt falling from heaven?||1|
|14||are you busy tonight?||1|
|15||You're such an interesting mystery!||1|
|16||What's your kernel version?||1|
|17||What are you doing here?||1|
|18||Wanna see my root file system?||1|
|19||My windows crashed today, how was your day?||1|
|20||Haven't seen you before...||1|
|21||Do you want to see my <geeky device>?||1|
|22||Do you run Linux, too?||1|
|23||Can I walk you home?||1|
|24||Are you tired because you've been running through my mind all night?||1|
It was amazing to see the big lead of dilbert. I'd expected to see much more userfriendly.org - and apparently Randall Munroe (xkcd) is quite popular, too.
|4||sound of heart||2|
|5||joy of tech||2|
|6||Fokke & Sukke||1|
|11||Kevin & Kell||1|
This was by far the most unusuable answer. Nearly everybody just said "penguin" or "tux". During the research for this question, I had already learned that the fire fox had no name - "Foxkeh" is just the name of the Mozilla Japan character. I also didn't know that a fire fox actually is a type of Panda - not a fox.
|4||the fire fox||2|
|7||Goi (squid, for webgui)||1|
Due to the fairly high number of MontaVista people interviewed, it could be expected that this question turned out skewed towards MontaVista. I also probably should have thought that most people wouldn't dare to mention another distro to my face. So, this question wasn't really useful and wasn't used in the game.
I'm not sure what happened here - I had one interviewee admit that he doesn't have a TV, and I had two candidates on stage who didn't own one. Anyway, I was amazed by the number of answers in favour of Friends.
|11||Two and a half men||1|
|13||Sex and the City||1|
|24||Buffy the Vampire Slayer||1|
The easiest question - and the only one that actually was completely successfully answered. I guess that if I'd added purple and bald, this question also would have failed on stage.
This question probably was a mistake. Coming from the Netherlands, where cable penetration is nearing 100%, I should have realized that this question would impose problems to the audience. Yet, most of my interviewees again had no problem, so only on-stage i realized that there actually are people without cable or satellite TV. Oops, my bad.
|3||how it's made||4|
Ford and Tata are seriously off - especially if one considers that none of the Indians I interviewed mentioned Tata - both occurrences were from Europeans who fancy their new extremely small car.
The first question trying to prove the regular computer geek food taste.
|6||Sisi / Fanta||2|
The second question trying to prove the regular computer geek food taste.
The third question trying to prove the regular computer geek food taste. This one actually is asked in one of the YouTube clips on "Family Feud". Milk was there too, the most popular answer. But most others were not present in their top3. I think geeks should clean their fridges more often. Also note the ice cream - apparently in some European countries a freezer is also called a fridge?
Back to the more serious topics here. Not surprisingly, "size" won.
|9||free / libre||2|
|10||flash file systems||2|
|14||protected memory model||1|
Not surprisingly, Linus won. Many people were surprised that Alan Cox actually still is mentioned that often, considering that he's no longer that visible. Apparently many interviewees approve what Ingo has done for the embedded world - he's fairly popular.
|13||Rik van Riel||1|
Deliberately, I had not asked for the favorite editor. However, I couldn't resist to bring up one of the other big controversies - scripting languages. People who know me, know that I'm not happy with the many choices for scripting languages here - I'd prefer if less scripting languages would be in actual use; who needs e.g. Ruby?
Note that many people mentioned different shell script types, I joined them all in the category 'shell'.
The final question. The number of mobile phones surprised me. I'd expected more laptops, computers and servers. But as most people don't buy these pre-installed *and* we're an embedded crowd, other devices won. Of course, by now the mobile phones are the most widespread amount of Linux devices - I guess there are more Linux phones than PCs+servers+desktops in the world.