The much misunderstood Ubuntu 9.10 upgrade poll

Gavin Clark at the Register recently reported that only 10% of people upgrading to 9.10 had a satisfactory experience. Serdar Yegalup at Information Week then reported that 40% of people upgrading to Ubuntu 9.10 had issues that they considered unfixable. Both of these conclusion were based on a poll on the Ubuntu support forums.

Serdar had the wisdom to point out that the poll is self-selecting (but still reported it). However, and I write this so you heard it here first, I don’t think we will have to wait long for a Microsoft shill to report his figure as fact and reference Information Week as evidence.

So let’s look at this poll. While we don’t have exact figures it is reasonable to assume that hundreds of thousands of people upgraded to 9.10 in the last few days. The quoted poll has a sample of 2158.

The poll is also on a support forum.

I upgraded to 9.10 a while ago. Flawlessly. So I saw little need to go tell a forum. This is where people go when they have problems. Gavin and Serdar were shocked to find people with support issues on a support forum. I have no doubt the help line at Microsoft has taken a lot of calls recently, but I would not extrapolate from that a large percentage of Windows users are having upgrade problems.

Tellingly and almost the last word on this are the polls from our previous releases, none of which were considered or reported as upgrade disasters:

Jaunty Poll

Intrepid Poll `

Hardy Poll

Gutsy Poll

A very useful summary of these findings by Nicholas Ipsen is here. I am linking to these polls not because I want to provide evidence that the Karmic upgrade experience is or was good or bad, there are other more qualified to comment on that, but that there is nothing new here.

All this of course is of little consolation if you are affected by an issue. Which is why we have the forums and Launchpad so that we can gather data, isolate the issues and fix them if they are an issue with Ubuntu or alert someone who can if they are not. We do this all the time with every release and we are doing it with Ubuntu 9.10.

So what we seem to have here is a poll that has existed for some 5 releases being ‘discovered’ and the data used to support a pre-disposed position. As we operate in the open and publish feedback good and bad, this is the risk we take I guess. Gavin and Serdar had some broader points that we could engage on regarding the readiness or otherwise of Linux for mainstream computing. But to base or support their arguments on this poll does little to illuminate and a lot to obfuscate.

Gerry Carr

Head of platform marketing, Canonical

7 comments

  1. Don Birdsall

    It is not true that only people with problems visit a support forum. If it were, then where did those people reporting a favorable experience come from? And where are those giving support coming from?

    As a a statistician (retired) , I am fully aware of the consequences of non-random samples. Mr. Carr contends that unsatisfied users are more likely to participate in the poll than satisfied users. There is no evidence of that and making that assumption is simply bad statistical analysis. The bottom line is that despite it’s possible inaccuracies, it’s still valid data and the only data we have. Mr. Carr needs to take a course in statistics before making further comments like this.

    If Canonical desires a better measure of their success they must create a random database of Ubuntu users and conduct independent unbiased polls. An (optional) registration of Ubuntu users would be a good start.

  2. Ben

    I see it this way also. Actually, my first upgrade attempt failed after my internet connection failed and I agreed to attempt a ‘Partial Upgrade’ after which I simply installed from the CD (after renaming my home folder and copying useful parts of the /usr/ resources) – There is always a sense of panic when going through the process. Having said this, a friend of mine upgrading to Windows 7 took several days to prepare before hand, writing stuff to a new 1TB External hard drive bought for the process – and he proclaimed it was flawless (though I saw him having trouble trying to get it working with his iPhone – he said this isn’t the OS, it’s an Apple problem)…

    Obfuscate is a good word, it is embedded not into Microsoft’s psyche, but also transferred to the mind of many MS users – who see a nightmare scenario out of which they emerge without losing both legs as a successful transaction. My previous experiences with Windows would suggest this is a very accurate view! It is, of course, completely incompatible with the real world, and cannot in any way be compared to alternative systems unless, perchance, the people involved have equal experience, bias and expertise on both sides of the fence.

  3. Endolith

    The real problem is that ALL the polls have at least 1/3 of voters experiencing major problems. This should be decreasing with each upgrade, not staying the same. I’ve had a graph of this going, too, which I think shows the trend more clearly: http://www.endolith.com/wordpress/2009/06/24/ubuntu-isnt-getting-any-better/

    I’m not sure I believe the poll is unrepresentative, either. For every customer who complains, there are 10 others who silently take their business elsewhere.

    I’d love to see a real polling organization do a neutral, unbiased poll comparing the ease of use of Ubuntu with other OSes.

  4. speedyx

    A marketing campaign for early-adopters that want to contribute in testing the next final release must be planned.

  5. donal

    Hey,
    I’m one of the “standard” Ubuntu users. Don’t compile, use CLI as cut & paste etc.
    But I’ve been on Ubuntu since 7.04, full time since 8.04.
    I stayed on Intrepid because of concerns about Intel graphics.
    I installed karmic, and voted on that poll after about 24 hours after installing.
    I gave it the positive “few small issues that i could fix” vote.

    Two weeks later I wish I could change that to serious errors I can’t fix. I wonder how many others like me have continually been finding problems:

    1: Wired Ethernet was fine but a week later when travelling I discovered Broadcom STA which had worked fine in Intrepid didn’t work. Activating STA only worked once. There are a lot Broadcom issues out there.

    2: Mobile 3G which worked fine in Intrepid also didn’t work. Again , a lot of people had this issue.

    3: External display, which had been fine in Intrepid, now requires me to run a script every time I reset to set the screens up the way they had been in Intrepid, with the External monitor as Primary, which seems to be contributing to a greater number of systems hangs. Again , others had this issue.

    4: Nautilus is hanging when viewing folders with video files. In Launchpad this is considered a “Low-Importance” bug!

    Other issues:
    5: Bluetooth Manger still can’t transfer/explore my phone.
    6; Lost GDM customising ability.

    5:

  6. donal

    Hit send before I was finished, but I’ve made most of my points.
    What bothers me most is the trend of things which previously worked, not working in this release. I’ve never contributed to any of the negative hysteria around each release, but I never had this level of dis-satisfaction before either.

  7. Alexavia

    I thought fiidnng this would be so arduous but itÂ’s a breeze!

Add your comment