On 3/28/07, nilfanion wiki <nilfanion(a)googlemail.com> wrote:
This stems from discussion in the Commons IRC channel
last night, and has my
own thoughts expanded somewhat.
Both of the procedures are somewhat broken, they basically come down to the
opinions of the reviewers. I feel the current system somewhat
disenfranchises people who are not regulars; both when nominating imagery or
when commenting on nominees. Part of this is it is *extremely vague* what
the two processes actually want.
Ignoring the difficulties that multiple languages bring lets compare the
English-language guidance for the three processes of interest to
English-language contributors: 1. Commons FPC, 2. Commons QI and 3. the
English WP FPC.
Firstly, COM FPC. It has two sections providing information to help people
inexperienced with the process:
[[COM:FPC#Guidelines for nominators]] and [[COM:FPC#Guidelines for
The first gives *very* brief details about what sort of imagery is expected
and the second drones on and on about technical details. This makes it hard
for new users to begin participating in either side of the process. A lot of
the votes are [[w:WP:IDONTLIKEIT]] or [[w:WP:ILIKEIT]], which isn't
objective - and leads to inconsistency (it depends who votes and what they
like). For example, images of animals taken in a zoo are frowned by some
regulars upon *merely* for being in a zoo. It is not clear from the criteria
why it is a big deal, or if it should be a big deal. Likewise, if you have a
stunning picture of an unidentified beetle, it will fail for being
unidentified - it is not clear from the guidance *why*.
Next, COM QI. This process is designed to be below FP - and is treated that
way. Unlike FPC it has a whole page giving useful guidance on the sort of
things that should be expected of a QI at [[Commons:Quality images
guidelines]]. These focus solely on the technical stuff, which seems
reasonable as QI is about the technical quality of the imagery first and
Finally, en FPC which I'm using this for comparative purposes here. Compared
to its counterpart on Commons, enwp FPC seems a much more hospitable
environment for new contributors. There is clear guidance on what is
expected at [[w:WP:WIAFP]]. The comments tend to be significantly more
wordy. Even more positively, when images show flaws regulars there will
often upload edited versions. Currently there are 46 images on COM FPC and
36 on enwp FPC. Over 12 of the en nominations have multiple images, but I
only count about 5 versions with edits on COM FPC.
My conclusion from all this is that enwp FPC is a more collegial process
than Commons. Of course its not perfect, but there are clearly lessons to
learn which we should apply. The fact it is English-only and we are
multilingual is a hindrance for us, but we can work with that. One thing I
noticed, and I think is bizarre is the following. On enwp FPC there is a
section [[w:WP:FPC#Is my monitor calibrated correctly?]] This gives
instructions to on how to carry out gamma correction; which is of key
importance before you can assess technical quality. COM:QI uses one of the
image on its guidelines page. COM FPC doesn't even mention it.
Actually, after an examination of what we "on paper" want from QIC and FPC,
I believe QIC has the more concrete definition and it is *harsher*. This is
clearly back to front. QI is the lesser of the two simply because of
inconsistency in review; its process is designed that way.
Anyway, that's my thoughts on the problems as they currently exist with
these processes. Now what to do about them?
Firstly, lets get Featured picture *criteria* analogous to en's WIAFP.
Obviously our requirements are different (encyclopedic value isn't a major
issue), but a set of concise, precise criteria makes it easier for all
people with any interest in the process. I'll expand on my thoughts as to
what we should include further on. The guidance for monitor calibration
should be copied across ASAP for the obvious reasons.
Secondly, reform is needed in how FP works. If the criteria become more
transparent and less precedent based, that encourages a broader community
giving input. Basically, we want positive feedback given and when a minor
fix is possible, for example reducing noise, ideally it should be carried
out by a reviewer; and the other voters would then review the edit. As well
as improving the atmosphere on the process - it will also improve the
quality of the output.
QIC is in better shape, but needs simplification and clarification. Is it
only about technical quality? If so it should say that. If not what else
matters? If it is only about technical content, should we really expect
everything to be *perfect* - what is reasonable leeway? How about some
enforcement of consistency allow a transparent method of review for
uploaders to say "I disagree with the review by X, I think it was a QI".
As for FPC: Here's my thoughts as to initial criteria - in no order of
1. Technical quality. Any FP should be able to pass the QI guidelines on
this front, and probably held to even tighter rules. It might be worth
explicitly stating low-resolution is bad.
2. Licensing. We don't actually *say* the image must be free. If we do, we
might actually get the reviewers investigating that properly...
3. A good caption: at a minimum the info needed for a POTD description. If
geo-data would be helpful, it should be included.
4. Value of image. We want to reward good images of difficult subjects and
we should probably give really hard subjects a bit of leeway on technical
quality to get it.
5. Neutrality and accuracy. This would be particularly relevant to maps and
similar images, but would be true for photographs too. In general, the image
should not promote a specific agenda or POV.
Commons-l mailing list
Thing is that QI was supposed to be a little less strict than FPC to
== Purpose ==
The purpose of quality images is to encourage the people (...)
While featured pictures identifies the absolute best of all the images
loaded into Commons, Quality images
**** sets out to identify and encourage users efforts in providing
quality images to Commons.***
Additionally quality images should be a place to refer other users to
when explaining methods for improving an image.
Yet QIC is very discouraging. IF an user doesn't like the image, they
get struck down, usually they point "it has X technical error" but
they don't suggest how to fix it /avoid it.
Notice that this "dislike" often has to do with artistic/composition
merits instead of techcnical details like Dschwen has pointed me out
It's like a roulette, you're up to the luck of the guy who reviews you
(unless you want to contest the decision and get into a review ).
Basically, QIC has become a "passes if has not a single techcnical
flaw" kind of thing, which isn't what it's meant to be. We have
guidelines, yes, but they're taken as word set on stone, and even
thought minor flaws could be overlooked according to criteria (well it
used to say so), common sense has gone away and as I wrote somewhere,
review has become just a checklist on techical aspects.
Now examples from current page:
=== Subjectivity examples===
:: Got approved: "Review DOF could be higher, but ear, eye, and
whiskers are reasonably sharp."
:: Some noise, extremely low DOF, eyes out of focus --Dschwen 07:41, 7
So, is low DOF good or bad?
::Composition good. Lighting bad - green colour cast (even a green
shadow on the monkeys arm), low light on face. Not quite in focus (or
is that camera movement - only 1/100th shutter speed)
Review I doubt the species identification, (...) I like hte composition
Hey, so, what about the techcnical aspects? Aren't those taken into
account? If tech aspects are important, why does it matter HERE the
taxonomy for animals, (it makes sense for FPC though)
: [APPROVED] Review Very good.
:::(More subjective I like it/ don't like it)
: [REJECTED] Review weak composition
:: ok, composition is important
: [APPROVED] Review Unfortunate framing, (...) Gets my blessing for QI though.
:: So weak composition is not important here?
: [REJECTED] Review Needs to be rotated counter-clockwise. (it's
about 2º tilted)
: [APROVED] Review The distortion is a bit disturbing, but apart from
that very nice.
: [REJECTED] Review Too much perpective distortion, easily fixed
: [APPROVED] Review Although not photographically perfect (some signs
of compression or diffraction), the composition and angle are
: [REJECTED] (from review) "but there is considerable chrominance
noise on the beach and there are precedences of QI declines for that."
So I'm getting a lot of mixed signals here, is a small tilt/distortion
enought to reject an image or not? IS a noisy image rejected or not,
since I see last two are similar wrt noise
: [APPROVED] Review I don't have the courage to say "no" to this guy...
A quote from Dschwen:
"Exactly, we have stricter rules in the technical department. QI
should be a reliable source for high quality images which are up to a
given technical standard. No matter how you turn it, this picture is
lacking in size and sharpness, especially if you consider that most of
it is dead area. FP is a different game, they are way stricter in
terms of subject matter and composition. --Dschwen 16:06, 28 March
So why some images which aren't photographically perfect are approved
and why others not?
And if I were not to assume good faith I would ask "why some regulars
are usually cut more slack than newcomers?"
I say the reason for this uneveness is that images are reviewed by a
single person, and it depends on the reviewer standards. In all case,
very discouraging, defeating the purpose of this, making people walk
away and turning this into WP:OWNED.
But before someone says, "Hey! you can appeal and there's a 2 week
process for that!" I have another batch of comments about the
reviewing process but I'll save those for the next time.
* Nomination Church in Göttingen. --Dschwen
* Review a clear image-LadyofHats 04:45, 26 January 2007 (UTC)
* For me - bad perspective. Lestat 19:50, 27 January 2007 (UTC)
o Please remember, that this is not FP. You are probably
alluding to the wide-angle effect. Whether this is bad is probably a
matter of personal taste. There are no obvious flaws such as tilt in
the picture. --Dschwen 18:56, 28 January 2007 (UTC)