Here is my answer to the statements made recently.
* For those who are not familiar with the issue, I am the main
developer of ProofreadPage, a Mediawiki software extension
that allows Wikisource users to proofread a page of text by
comparing it to its scanned source. This extension also manages
book metadata and citation information, and it imposes a very basic
processing workflow, where a page must have been checked by two
different users in order to reach its final state.
[see here :
http://en.wikisource.org/wiki/Help:Page_Status]
* This extension has been introduced four years ago. Since then, more
than 350000 pages have been proofread with it, and more than 120000
pages have been double-checked. By now, between 500 and 1000 pages
are being proofread every day at Wikisource. These figures are
growing rapidly, and we can reasonably expect that Wikisource will
become more active than Project Gutenberg's Distributed Proofreaders
in the coming years.
* A few de.ws admins have requested that the ProofreadPage extension
be modified, in order to allow anonymous users to mark pages as
"proofread". It is very important to understand that they do not
request only that : They want any user, anonymous not, to be able to
set any page to "validated" (quality level 4), no matter if it has
passed through the "proofread" stage (quality level 3). And this is
indeed what would happen if IPs were allowed to mark a page as
"proofread". This would break the "two proofreaders" rule.
* As a matter of fact, these de.ws admins are not opposed to the
"two proofreaders" rule. However, they want this rule to be
enforced by themselves, not by software. Indeed, before the
introduction of ProofreadPage, the "two proofreaders" rule was
enforced at de.ws by the community, and this rule did not exist at
other wikisources.
* The "two proofreaders" rule that is built into the software is
not meant to be robust ; it is very easy to circumvent by using
sockpuppets. And its goal is not to prevent vandalism, as is
sometimes stated. No, the goal of this software restriction is to
ensure that various users, who do not necessarily read the rules,
or who do not interpret them in the same way, share a common
interpretation of the quality levels. Its goal is to make their meaning
unambiguous.
* I do not think that the "two proofreaders" rule can be enforced
in the long run by a wiki community, without software. This rule
has been enforced at de.ws for some time, because it was a small
community, where most users knew themselves, and where rules were
very strictly observed (my goal is not to spread stereotypes about
the Germans; they are very proud about their superior quality
standards). However, such a level of organization and rule
enforcement cannot be achieved at other wikis, because they are much
more lax and larger communities, with less rules, with users who do
not take time to read the rules, with users who sometimes disagree
with rules, and where most users do not like to play the police. In
the long run, enforcing that rule will become impossible at de.ws
too.
* In a lax community, where users do not always read the rules before
participating, if a rule is not clear or ambiguous, then users start
to develop their own interpretations of it. In this context, if any
user is allowed to set any page to quality level 4, no matter if
that page has been proofread before, then users will start to have
diverging interpretations of the meaning of q3 and q4. For example,
some of them will decide to use q3 for proofreading, and q4 for
formatting. This is not a thought experiment: It already happened at
fr.ws, with our previous proofreading system based on icons. And you
cannot blame users for that : it is very intuitive to think that q3
means "unfinished", when there is another level called "q4".
* In contrast, if software does not let you reach q4 but only q3, no
matter how well you proofread and format a page, then it does not
make sense to believe that you are supposed to leave some part of
the formatting work for later, for the person who will be doing
q4. The only interpretation that makes sense is that you should do
as much proofreading and formatting work as you can. And if you
use a sockpuppet in order to reach q4, then you _know_ that you're
doing something you're not supposed to do.
* Thus, removing the "two proofreaders" rule from
ProofreadPage would spell the end of the current proofreading
workflow. Again, this is not a thought experiment, but something
that already happened at fr.ws with our previous system. Once a wiki
decides to enable such an option, we will see users making diverging
interpretations of the quality levels, and the pages that were
previously validated by two users will be in the same category as pages
validated by a single user. This would be a complete lack of respect for
the validation work that has been accomplished so far.
* For this reason, I will never allow such a change to be made to
ProofreadPage. I do design and maintain this extension by listening
to the requests of users, and I do implement most of the features
they request, or add patches submitted by other users. However, when
a request is so blatantly contradicting the purpose of the tool, I
believe that I have the right to refuse it.
* Please note that I am not opposed to IPs, and my goal is not to
discriminate them. And I would certainly allow IPs to mark pages as
"proofread", if we could find a way to do this in a way that does
not hurt the "two proofreaders" rule. I even proposed to allow pages
status to be modified by a whitelist of static IPs, but my proposal was
ignored.
* As a matter of fact, there are various other things that anonymous
users are not allowed to do at the Foundation. One restriction that
is particularly relevant for Wikisource is the fact that IPs are not
allowed to upload scans. This implies that they are not allowed to
start a new proofreading project. Strangely, this intolerable
discrimination does not seem to hurt anyone at de.ws. If their fight
is really about the rights of IPs, why don't they complain about
that?
* More generally, I do not feel comfortable when someone pretends to
speak for a category of people who do not express themselves. When a
minority is silent, you can put any words in their mouth. According
to Michael Joergens, these anonymous users are scared of getting
registered and of giving away their email addresses, and he is mandated
to speak for them. But how can we check their existence, how can we
check how many they are, if they refuse to be identified ?
Finally, several statements have been made in the previous messages,
that are wrong or misleading :
* Firstly, it was written that de.ws _must_ use the ProofreadPage
extension. This is not true. There are other similar proofreading
tools that have been developed at de.ws, and these tools are still
used today. You are free to use or to develop whatever you want.
* Secondly, it is false to claim that there is a consensus at de.ws, in
favor of allowing IPs to mark pages as "Proofread". The statement
made in this mailing list, about the existence of such a consensus,
has been questioned shortly after in the de.wikisource Scriptorium.
So, if Klaus Graf wants to speak in the name of the de.ws community,
he should ask them first. Eight months ago I suggested that a vote
be organized at de.ws, on that question; instead of that, the same
group of admins held a vote in order to have me desysoped (and they
lost it).
* Thirdly, if the de.wikisource community decides, by vote or by
consensus, that they want IPs to be allowed to change the quality
status of pages, they can do this without destroying
ProofreadPage. They just need to step out of the ProofreadPage
quality system, and restore their previous system in place of it.
I am willing to explain how to do this to any technically skilled
person. Note that I already made the same proposal in bugzilla
8 months ago.
I apologize for the length of my answer. I wish to thank those who
have had the patience to read this entire post.
Thomas