Linux Magazine Criticizes Karmic

[Samuel Thurston, III]

On Fri, Nov 6, 2009 at 8:05 PM, Florian Diesch <diesch at spamfence.net> wrote:
> To goal is to compare different versions of Ubuntu, not to get absolute
> numbers. And I think it would be at least surprising if 9.10 was the
> worst release so far but the polls show it as the one with the least
> problems.

Sure, and I'm not saying it's a pointless effort..  it's just that the
line quoted: "if you do take these polls at face value" contrasts
rather starkly with his intro: "they can't just be taken at face
value."  I found it humorous that's all.  If they weren't making some
kind of internal progress I'd be horrified.

> In a support forum I'd expect to find more people with problems they can't
> solve themselves than people who don't have any problem.

Well that depends on perception: if you ask a question in a forum and
it leads you to figure out the problem, did you solve it yourself?
I agree that you're more likely to have people with some kind of
problem on the forum in the first place. The question is, how many who
can't solve the problem will stick around to answer a survey?

You can find it quoted many times in marketing literature (however I
don't know the original source or if it's in fact accurate) that only
4% of dissatisfied customers ever complain while 96% disappear
silently to the competition.  It's logical from a common-sense
perspective, which is why I doubt its accuracy.

>
>> there is only a 28% success rate, and that presumably includes some
>> experienced users coming in from previous releases.  That's abysmal.
>
> Well, if you want to take the raw numbers serious:
>
> That's about 58% of the people who did a new install. Updating worked
> for about 46% of those who tried it.
>
> About 66% of all did successfully get 9.10 (either by a new install or
> by upgrading).

Bugger.  That's what I get for not reading the numbers and getting out
a calculator.

>
>
>>  Less than a third of people trying Ubuntu
>> for the first time are able to successfully install it?
>
> No. 100% are all the answers, not all the people who tried a new install.
>

In that event all the percentages are useless for comparison since
you'd need a baseline of people upgrading versus installing.  I mean
it's great that he's comparing the surveys but he obviously doesn't
understand statistics.

Rebasing _installs_ only at 1048 (just for 9.10,) 384  (36.6%)
reported "many problems i've not been able to solve" meaning 63.4%
succeeded in getting it working.  Ok so that's less than 2 out of 3.
Not as bad as I had thought before but still really horrible,
especially given the self-selection bias.