Hey, so I’ve been pondering the results of the ‘getting Fedora’ survey that I compiled last night. Here’s what I’ve been thinking about and some of my ideas. Please share your own and let me know if I’m off-base with any of my ideas 🙂
- The vast majority of survey respondents came from planet.fpo and planet.GNOME, so we can assume that this survey represents responses from users who are experienced with open source & Linux and are likely quite techincal.
- The number one trigger by far (a trigger for 92% of respondents) for downloading a Fedora release is when we have a new release. This means a few things – download volume and the visibility of our download pages will be highest around a release. If we want to make a high-visibility change, have the greatest impact, and get the highest volume of feedback on changes, making the changes around a release is ideal. If we want to make a more experimental change and don’t want to affect a lot of people right away and go about things a bit more quietly, inbetween releases is best.
- New hardware, friends, and hardware failure are are the 2nd, 3rd, and 4th reasons respectively that people download Fedora. These were almost always listed as a secondary reason to download Fedora. Fewer respondents indicated downloading Fedora in order to upgrade a system whose Fedora version was no longer supported. I think this probably indicates a lot of the respondents like downloading and installing the latest rather than hanging back on older releases in order to maintain stability.
- Just an idea: we don’t advertise Fedora as a solution to help diagnose/fix problems with other OSes or HW issues, yet they do seem a common trigger for downloading Fedora. To reach out to these users with some documentation might be an interesting avenue to recruit users from other OSes.
- 46% – less than half – of respondents visit www.fedoraproject.org to obtain a copy of Fedora. This is actually a lot less than I was expecting. The 2nd, 3rd, and 4th reasons, which are all used by about the same percentage of participants (15-19%) are going directly to torrent.fpo, preupgrade, and going directly to get.fpo. We have a LOT of different methods for getting Fedora… and besides www.fpo their usage seems scattered. I think there are many users who know what they want (torrent vs mirror vs spins vs etc etc etc) so it’s important that we maintain and socialize a consistent and easy naming scheme for these various download portals so folks who know what they’re doing have an easy URL to remember and punch in. Many of the portals (mirrors.fpo is an example) aren’t really branded with what they are, so branding these sites with clear names and pushing their URLs maybe even within the body of the page might be an effective way to make them easier for these users to remember.
- Spins.fpo is not currently widely used, with only 6% of the 78 respondents reporting having using it. For us to rely on it more in the future means we definitely need to plan on socializing it and marketing it widely.
- Experienced users have a lot of different ways they download Fedora, and many do not involve the website – local mirrors, PXE, FWN/Announcement links, even their own mirrors and surprisingly using yum to upgrade. If we target experienced users for downloads on www.fpo, more than half of them aren’t even going to bother to look.
- Users from other distros do download Fedora to take a look but not necessarily to migrate. A few mentioned trying it out on virtual machines. The release notes are probably important to these folks so they can see if there’s anything new that they want to check out.
- By far the two most popular downloads of Fedora are the DVD (63% of respondents)and Live Media (49% of respondents.) To me this indicates that the DVD shuld probably be the heaviest-weighted option on the screen(s) we put together for more advanced users.
- There are respondents who indicated wanting a copy each of GNOME and KDE live media, so on the spins page these should be labeled as such (e.g., the Desktop spin should have some indication that it is GNOME-based.)
- Boot-image/netinstall/etc and preupgrade were Fedoras respondents ‘got’ and had respectable numbers – 22% and 21% respectively. Preupgrade is getting a lot of usage despite not being advertised on the current download pages.
- Several respondents pointed out they prefer to download the minimum (boot image / initrd / etc) and download only what they need on demand so as to not waste space on things they never use. It would be interesting to dig a little bit more into this case and consider tailoring to these folks on the advanced page as well, as it seems 1 out of 5 respondents use this minimal install method.
- Not many respondents indicated relying on purchasing physical media or the Free Media program, which makes sense since if they have low-no internet connectivity, they couldn’t have answered the survey. 🙂
- Many respondents indicated they use the LiveMedia either strictly or primarily as a ‘showcase,’ so it may be worth considering posing it as such.
- Several folks indicated that they install from Beta or Prerelease media and track upgrade from there. This is an option not detailed or outlined anywhere. One respondent actually stated they wanted to help out by testing things and such by using rawhide and didn’t know how to do this. This may indicate a need for better documentation on what ‘Rawhide’ is, why you would use it, and how to get it and use it.
Would it be worth reformulating this and trying to get it in front of a broader audience, like fedora-list, lwn, etc.?
Probably, but at the same time – I think we have a lot of responses from experienced technical folks. I think the greater need is to do a survey of FOSS/Linux newbies to understand how we can attract them if we so desire.
If I was gonna do a wide distribution, I'd want to do it right – there were some choices I didn't think of that I should have provided based on the responses, and also counting everything manually is painful and error-prone. 🙂
Yeah, I just figure planets in particular are super-technical/super-educated; lwn/fedora-list are not ideal but at least broader than the planets. I wish I had a better suggestion- perhaps reach out to a tech site like arstechnica which presumably has technical users who have experimented with Fedora, but who aren't necessarily full-time users?
As far as 'doing it right', you can use Google Docs to really easily generate web surveys and collect the data, should you choose to do it again. (I figured you'd have learned about the answers- that is why I suggested it.)
Shame there aren't the resources (AFAIK) to do really high quality analysis of download data like http://blog.mozilla.com/metrics/2009/01/07/a-user…
Only 6% using spins.fp.o, may mean only 6% of people are interested in getting those spins, which seems logical, as all the spins are somewhat 'niche'.
[…] my $0.02 on the 'getting Fedora' survey results http://mairin.wordpress.com/2009/08/27/getting-fedora-survey-result-discussion/ […]
> * This may indicate a need for better documentation on what ‘Rawhide’ is, why you would use it, and how to get it and use it.
This page is a very good base: http://fedoraproject.org/wiki/Rawhide
Advertising it on the download page (for experienced users only?) could be a good idea.
A couple of things to consider, since releases are the major trigger for downloads:
– In the past, some Fedora builds haven't been very VMWare-friendly. I know of plenty of geeky folks who are abandoning their home "datacenters" with a real beefy box running a hypervisor. At my job, we're starting to replace lab gear with VMWare… if something doesn't work well right out of the box, we go elsewhere.
– It might be a good idea to figure out how to move away from the Monolithic DVD model towards a more lightweight model.
I think the second point is important, because many people think that apt is some special magic and Ubuntu/Debian are the magicians. That was true a decade ago, but yum and other tools have similar functionality, but many people don't know about it. Those people may be without clue, but new users tend to be that way.
Seeing a strong majority prefer the install DVD, I would argue the other way. I would even go further: the larger 64bit DVD ISO is about 3.5 GB, we should help this majority of users and include *more* software on it (towards 4-4.2 GB) and not wasting the space.
As for you versus apt-get, what can I say more? DeltaRPMs, this totally kick ass.
Thanks for taking the time to quantify the data and do the analysis.
[…] Survey Results Discussion […]