On 03/08/2010 03:31 PM, Adam Williamson wrote:
> On Mon, 2010-03-08 at 12:14 -0500, Doug Ledford wrote:
>> On 03/08/2010 11:05 AM, Adam Williamson wrote:
>>> If you think the poll is wrong - provide some data to disprove it.
>>
>> I'm sorry, but that's a scientifically specious argument. Invalid data
>> doesn't become valid because there is no valid counter data. It is
>> valid or invalid all on its own. To date, no one has run a
>> scientifically valid poll, but that doesn't make your poll any better or
>> worse, it just makes it all by itself.
>
> My basic point here is that the poll, while imperfect, is the best
> indication we have available so far.
So? From a scientific process perspective, bad data is bad data. And
if all you have is bad data, then you really have no data at all. There
is no such thing as "the best we have available" translating to "not
bad" or to "usable".
> My second important point is that complaining about a poll being
> problematic and backing up your complaint with nothing but utterly
> unsupported assertions is entirely hypocritical. If my data is invalid,
> their assertions are...well, even *more* invalid (although validity
> isn't an analog concept, I accept).
You have your roles backwards here. In the realm of scientific
scrutiny, it is up to the researcher to be able to establish beyond a
reasonable doubt to their critics that their methods do in fact meet the
necessary standards for scientific integrity and that the
research/experiment does indeed prove what they try to establish it
proves. In other words, it's up to you to defend your poll if you want
it to be taken as though it is in any way authoritative or accurate. It
is not up to the people criticizing your poll to prove their assertions.
The burden of proof is all on you.
>>> Counteracting it with yet more assertions built on precisely no evidence
>>> is not convincing.
>>
>> Well, one of the questions to be asked before going any further on this
>> is what audience do we care about? I've heard it over and over again
>> that Fedora is supposed to be a developer's platform, and not a user's
>> platform. If that's true, then the people that should be voting on this
>> is the people that make Fedora, not the people that consume it. If the
>> reverse is true, then it really doesn't matter what the users vote
>> anyway because then it's up to us to decide *which* user segment we wish
>> to target and build the OS to satisfy them.
>
> I agree, and I'm one of the people who's been saying this for months. On
> a practical level, though, it doesn't look like it's going to happen any
> time soon, whereas by some of the comments in this thread, some people
> seem happy to claim that FESco has sufficient authority to decide an
> updates policy on its own, and also seem as if they have some
> inclination towards doing so.
>
> So it seems like there may be efforts to make changes in the update
> policy situation _before_ the target audience issue is settled (which,
> like you, I do not think is a good idea).
Good, I'm glad we agree on something ;-) Pick your target audience,
then make things work for that audience.
> As I said right back at the start, I primarily did the poll because more
> than one person in the thread happily asserted that 'users don't want
> adventurous updates', without bothering to provide any kind of support
> for that claim. It's a lot harder to make that claim now, I think. Even
> if you want to argue that the poll isn't sufficiently rigorous to
> 'prove' that users want adventurous updates, I think it's sufficient
> data to make it clear that barely asserting that users don't want such
> updates isn't admissible.
>
>> Now, as for the wording. It was both subjective and vague. Neither of
>> those leads to a good poll without at a minimum putting in additional
>> questions to narrow down responses. As an example of why I call it
>> subjective and vague, I could have worded the same "adventurous" and
>> "conservative" options as "gratuitous" and "reasonable",
>
> That's not a very good example, because you're taking the wording that I
> claim is good and replacing it with bad wording and saying 'this proves
> the wording is bad'. Huh?
No, I'm taking wording that I claim was bad and replacing it with
wording that was equally bad in order to demonstrate the point. Just
because "adventurous" and "conservative" are not loaded and subjective
terms in your mind does not mean that they aren't in other people's
minds. The point of wording in a poll like this is that the choice of
wording needs to be neutral in as many people's minds as possible.
Moreover, adventurous and conservative are too vague to be useful. Does
adventurous mean that you take the latest version from upstream
immediately on release every time and dump it straight into updates?
Does it mean you go a step further and pull direct from the upstream
repo and don't even wait for a release? Maybe you go even further and
instead of pulling from the official upstream repo you pull from the
equivalent of the for-next kernel repo? Likewise, conservative could
mean nothing bug security/serious bug fixes, it could mean minor point
updates, in contrast to pulling from for-next it could even include
major point updates as long as they are officially released by upstream.
So I stand by my claim that adventurous and conservative are both loaded
terms (maybe not to you, but certainly to some people, myself included),
and I stand by that they are too vague to be used in a single question
poll and have them produce any meaningful results.
> By that token you could take any well-worded poll, replace it with bad
> wording, and say 'the fact that I can plug bad wording into this poll
> question means the original wording must also be bad!' It's a
> non-sequitur.
Sorry, it wasn't well worded. That you think it was is part of the
problem and why I gave you the demonstration I did.
> The whole reason I chose the words 'adventurous' and 'conservative' is
> that I don't believe either of them have especially positive or negative
> connotations. Which is clearly not true of 'gratuitous' and
> 'reasonable'.
They most absolutely have connotation.
>> To be a valid poll, we have to be precise
>> enough that people know what they are voting on without the wording
>> leading their thoughts.
>
> In theory I agree. In practice, it's impossible to ensure this (_any_
> poll question can be deconstructed), and I don't think the wording I
> chose was significantly leading.
I disagree.
--
Doug Ledford <dledford@xxxxxxxxxx>
GPG KeyID: CFBFF194
http://people.redhat.com/dledford
Infiniband specific RPMs available at
http://people.redhat.com/dledford/Infiniband
Attachment:
signature.asc
Description: OpenPGP digital signature
-- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
