Open Letter: Why I, Kevin Kofler, am not rerunning for FESCo

[Przemek Klosowski]

On 05/03/2010 10:01 PM, Jesse Keating wrote:
> On Tue, 2010-05-04 at 01:58 +0200, Kevin Kofler wrote:

>>> It was none of that.  All it gave us was info we already had.  Some users
>>> would like more adventurous stuff, while some users would not.  We already
>>> had that information, the poll told us nothing new.
>>
>> Now you may try to argue that the sample is biased, but you have no actual
>> evidence towards that.
>
> Of course the sample is biased.  It's a sample of people who frequent
> the forums, that's a self selecting group of people, by no means a
> worthwhile representation of the Fedora user base as a whole.

Besides the statistical bias, I think this poll is flawed because it is 
sensitive to how the issue is worded, and even how people perceive the 
question; it's like those 'push polls' in politics. Consider those three 
formulations (*):

  - would you like more adventurous stuff in Fedora, to take advantage
    of improvements and fixes in the installed software

  - would you like more adventurous stuff, even if it sometimes
    introduced regressions

  - knowing that sooner or later it will totally break your system,
    would you like more adventurous stuff

The results will be different for each question. They will also strongly 
depend on the cohort you will be asking. I am convinced that general 
users will be more conservative than the Fedora developers, who might 
answer 'yes' even to the third question, because they know enough to 
have a fighting chance to recover their systems.

An average user wants a stable, working software distribution, with
prompt patches and software enhancements. Since in general those are 
conflicting requirements, the Fedora community has to apply engineering 
judgement on what is the appropriate velocity of updates. I personally 
like the processes and infrastructure that Fedora built to manage the 
updates, even though I have seen that they don't prevent broken software 
from getting in. My conclusion is to make it better,
not jettison it.


p


(*) I tried to bias the wording to show the range of possible 
interpretation, so please don't call me out on trolling. The third 
formulation, while admittedly alarmist, isn't completely unrealistic: 
c.f. the recent scary bug with a localization interaction that ended up 
removing large amount of packages, including yum IIRC.