The author attribution survey is now closed. We have 1017 complete responses. I've posted results of the attribution data in the following report:
http://wikimediafoundation.org/wiki/File:Attribution_Survey_Results.pdf
I've posted the raw data of the attribution survey here:
Respondents from English Wikipedia: http://wikimediafoundation.org/wiki/File:Attsurvey-en.ods Respondents from German Wikipedia: http://wikimediafoundation.org/wiki/File:Attsurvey-de.ods Respondents from miscellaneous languages and projects: http://wikimediafoundation.org/wiki/File:Attsurvey-misc.ods
(The survey was linked via the WikimediaNotifier bot, so we got quite a bit of nicely dispersed traffic.)
As the report shows, and as I indicated in my prior e-mail, there is wide support for simple attribution models, and fairly strong and visible opposition to full author attribution (as well as complete absence of any attribution). Full author attribution is the second least popular option, at 32.82%. Many comments pointed out the tension between free content and attribution, such as:
* "While the whole point of Wikipedia is to provide access to information freely and easily, a balance must be struck between recognising authors' contributions and the constraints on utilising the information." (User's preferred attribution model is link to the article.) * "Giving credit to all authors is ridiculous! I think the 'Wikipedia Community' is sufficient credit, this project is not about personal gratification, its about community collaboration." * "Full list of authors is terribly impractical." * "Including the full list of authors on a 'NOT online' resource would be a waste of resources, i.e. paper and ink, most of the time. But even for online use, who would read the version history? On the other hand, a link can't do much harm..." * "Establishing which editors to credit would cause enormous disagreement" * "Although requiring credit may sound noncontroversial, it actually is a pretty big can of worms in contexts of (a) editing wikipedia-sourced content into rather different things (for example, the way that some wikipedia articles grew out of 1911 Britannica articles), (b) what if the wikimedia foundation has some kind of meltdown and it is necessary to fork the project. Therefore my recommendation is to not think in terms of 'requirements' but suggested practices."
Some users commented on the fact that Wikipedia is primarily written by people under pseudonyms, and that being suddenly visibly attributed would actually come as a surprise:
* "If any version of credit-sharing citing editors is made policy, all editors should be given notice and allowed to change their monikers to their choice. In my case, I choose a moniker I liked when I thought the community would remain anonymous forever. If my contributions went into print or were used similarly I would like to use my actual name."
Community credit proved a quite popular option, second only to a direct link to the article. Many people viewed it as a simple method to credit their contribution both online and offline. (At least one user suggested linking to detailed histories online, and crediting the community collectively offline.)
A few users felt very strongly about always giving author credit. The strongest example I found:
"I won't accept nothing less than what I chosed above, and I'm ready to leave my sysop status and other wmf-related roles if WMF will underestimate the meaning of GFDL to our projects. GFDL is what we would have chosen if asked 8 years ago, and is what we will stand up for."
Some users also pointed out that our options were constrained by the requirements set forth in the GFDL.
I'd love to see deeper analysis of the survey. I want to restate my original intent in running it: it's intended to be a feeler survey, to get a rough impression of what attribution models are widely considered acceptable by contributors to our projects, and which ones aren't. It served this purpose, and I have no intent in running additional surveys; we're on an aggressive timeline and have to move forward. It's also not intended to dictate a solution.
My preliminary conclusion is that a simple, manageable attribution model, while causing some short-term disruption, will widely be considered not only acceptable, but preferable to complex attribution models, in support of our mission to disseminate free information. That being said, we probably still have to find a compromise, as well as language that appropriate deals with single-author multimedia contributions. I imagine that if we a) have a more prominent "list of authors / list of people who contributed to this revision" credit link on article pages; b) require that a link must be given, and that the preferred linking format is to the revision that is being copied, c) explicitly state in our attribution terms that for images, sounds and videos that aren't the result of extensive collaboration, credit must be given to the creator, we're covering most cases.
We then still have to resolve the issue of giving credit for content imported into our projects consistently, which is a bit of a can of worms. (We might want to set some limitations on what kinds of content we import, to prevent "attribution pollution".) But it's secondary to the main issue of a consistent attribution model within our projects.
A model like the above is consistent with CC-BY-SA. There is a question as to whether it can be reconciled with our current practices. I believe it can, and I also think we can find mitigation strategies for contributors who vehemently disagree. I'll work on a revision to the currently proposed language, and will post that next week, alongside some further thoughts.
In terms of our timeline, I don't believe we can wrap things up prior to the Board meeting in April, but I think we can still hit a timeline to make a migration decision by mid-to-late April. SPI has committed to help administer the vote as an independent third party. What still needs to be done:
* We need to form a little workgroup/committee to help with the usual process of tallying the votes; * We need to translate all relevant text (including the vote announcements), once it's final, into as many languages as possible; * We need to implement a modified Special:Boardvote so it can be used for this decision. * We also want to allow sufficient time for the actual decision-making, ideally 3-4 weeks.
We have a big all-staff meeting and an all-day tech meeting next week, which will hamper us a bit in moving this forward aggressively, but I'll see if I can move things along a bit before then. If someone wants to create draft pages for any of the above (workgroup, announcement, etc.), I'd be very grateful :-)
More soon, Erik
2009/3/7 Erik Moeller erik@wikimedia.org:
The author attribution survey is now closed. We have 1017 complete responses. I've posted results of the attribution data in the following report:
http://wikimediafoundation.org/wiki/File:Attribution_Survey_Results.pdf
I've posted the raw data of the attribution survey here:
Respondents from English Wikipedia: http://wikimediafoundation.org/wiki/File:Attsurvey-en.ods Respondents from German Wikipedia: http://wikimediafoundation.org/wiki/File:Attsurvey-de.ods Respondents from miscellaneous languages and projects: http://wikimediafoundation.org/wiki/File:Attsurvey-misc.ods
Thanks.
I'm curious, why did you include options that aren't actually available? No credit and credit to the community are clearly not in keeping with the license, so knowing who would accept them isn't particularly useful (although I'm not sure it hurts).
Baseline, maybe?
________________________________ From: Thomas Dalton thomas.dalton@gmail.com To: Wikimedia Foundation Mailing List foundation-l@lists.wikimedia.org Sent: Saturday, March 7, 2009 10:18:01 AM Subject: Re: [Foundation-l] Attribution survey and licensing next steps
2009/3/7 Erik Moeller erik@wikimedia.org:
The author attribution survey is now closed. We have 1017 complete responses. I've posted results of the attribution data in the following report:
http://wikimediafoundation.org/wiki/File:Attribution_Survey_Results.pdf
I've posted the raw data of the attribution survey here:
Respondents from English Wikipedia: http://wikimediafoundation.org/wiki/File:Attsurvey-en.ods Respondents from German Wikipedia: http://wikimediafoundation.org/wiki/File:Attsurvey-de.ods Respondents from miscellaneous languages and projects: http://wikimediafoundation.org/wiki/File:Attsurvey-misc.ods
Thanks.
I'm curious, why did you include options that aren't actually available? No credit and credit to the community are clearly not in keeping with the license, so knowing who would accept them isn't particularly useful (although I'm not sure it hurts).
_______________________________________________ foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
2009/3/7 Thomas Dalton thomas.dalton@gmail.com:
I'm curious, why did you include options that aren't actually available? No credit and credit to the community are clearly not in keeping with the license, so knowing who would accept them isn't particularly useful (although I'm not sure it hurts).
We tried to surface people's "true preference" for an attribution model. (While of course the provided options can't capture everything, the relatively low number of write-in options for additional attribution models suggests that respondents generally found their views represented somewhere in the continuum of given options.) People's true preferences should guide our thinking process, and if we clouded the available options with perceived or real constraints, we wouldn't be able to approximate the best feasible solution. It helps us to uncover both where people may be willing to compromise and where they may not be.
For example, if the survey had shown community credit to be highly desired and not controversial at all, that would be interesting: We could have an informed conversation about whether we should try to accommodate that model after all. As it is, it's the second most popular first option, but with 15.29% ranking it as their second-to-last option, it's also somewhat polarizing. A link to the article, on the other hand, is the first or second option for more than 60% of respondents, and the last or second-to-last option for only 3.47%.
On Sun, Mar 8, 2009 at 4:46 AM, Erik Moeller erik@wikimedia.org wrote:
For example, if the survey had shown community credit to be highly desired and not controversial at all, that would be interesting: We could have an informed conversation about whether we should try to accommodate that model after all. As it is, it's the second most popular first option, but with 15.29% ranking it as their second-to-last option, it's also somewhat polarizing. A link to the article, on the other hand, is the first or second option for more than 60% of respondents, and the last or second-to-last option for only 3.47%.
Erik, this is not relevant because two options were non-options, so in the opposite case third-to-last is also relevant. (Even it was good to include them to realize how much Wikimedians care about their authorship.)
On Sat, Mar 7, 2009 at 11:32 PM, Milos Rancic millosh@gmail.com wrote:
On Sun, Mar 8, 2009 at 4:46 AM, Erik Moeller erik@wikimedia.org wrote:
For example, if the survey had shown community credit to be highly desired and not controversial at all, that would be interesting: We could have an informed conversation about whether we should try to accommodate that model after all. As it is, it's the second most popular first option, but with 15.29% ranking it as their second-to-last option, it's also somewhat polarizing. A link to the article, on the other hand, is the first or second option for more than 60% of respondents, and the last or second-to-last option for only 3.47%.
Erik, this is not relevant because two options were non-options, so in the opposite case third-to-last is also relevant.
If you want to draw the line at "controversial", fourth-to-last is also relevant, since "link to history" is nearly as controversial as "link to article". Of course, then you have a mixture between people who intended to rank "link to article" as third best, not fourth worst. You really can't draw conclusions from anything other than the pairwise results (and, like I said, even that is potentially misleading if people want multiple forms of attribution, such as a link to the article *and* a list of authors, or credit to Wikipedia *and* a list of authors).
I guess it'll become more clear when the binary decision comes up for vote: GFDL, or link to "any transparent copy that includes the same licensing and authorship information as the Wikipedia.org" (which wasn't even one of the choices), presumably. My guess, looking at the heads-up results, 80/20 in favor of the later. If it turns out to be closer to 96/4, I'd be quite surprised.
2009/3/8 Milos Rancic millosh@gmail.com:
On Sun, Mar 8, 2009 at 4:46 AM, Erik Moeller erik@wikimedia.org wrote:
For example, if the survey had shown community credit to be highly desired and not controversial at all, that would be interesting: We could have an informed conversation about whether we should try to accommodate that model after all. As it is, it's the second most popular first option, but with 15.29% ranking it as their second-to-last option, it's also somewhat polarizing. A link to the article, on the other hand, is the first or second option for more than 60% of respondents, and the last or second-to-last option for only 3.47%.
Erik, this is not relevant because two options were non-options, so in the opposite case third-to-last is also relevant. (Even it was good to include them to realize how much Wikimedians care about their authorship.)
Indeed - you need to be very careful when analysing survey results. Including non-options does provide extra information, but it means you need to take care how you interpret things. Those two options ought to be removed before calculating percentages in the way Erik has done.
On Sun, Mar 8, 2009 at 9:24 AM, Thomas Dalton thomas.dalton@gmail.comwrote:
2009/3/8 Milos Rancic millosh@gmail.com:
On Sun, Mar 8, 2009 at 4:46 AM, Erik Moeller erik@wikimedia.org wrote:
For example, if the survey had shown community credit to be highly desired and not controversial at all, that would be interesting: We could have an informed conversation about whether we should try to accommodate that model after all. As it is, it's the second most popular first option, but with 15.29% ranking it as their second-to-last option, it's also somewhat polarizing. A link to the article, on the other hand, is the first or second option for more than 60% of respondents, and the last or second-to-last option for only 3.47%.
Erik, this is not relevant because two options were non-options, so in the opposite case third-to-last is also relevant. (Even it was good to include them to realize how much Wikimedians care about their authorship.)
Indeed - you need to be very careful when analysing survey results. Including non-options does provide extra information, but it means you need to take care how you interpret things. Those two options ought to be removed before calculating percentages in the way Erik has done.
Still wouldn't be valid: it is quite reasonable that someone who finds linking unacceptable ranked "link to article" third, with "link to history", "no credit", and "community credit" behind it (or not ranked). There appear to be 9 cases of this ranking in the english data - 1.5% (whether they found linking "unacceptable" or merely "less acceptable" cannot be determined from the data given).
It's also reasonable someone ranked link to article second even though they found it unacceptable. "Full list of authors must always be copied" is really a non-option as well, since it contradicts the current practices of nearly all reusers. "Online: link; offline: list authors" contradicts the current practice of very few reusers - in fact the only noteworthy one I'm aware of (Wikipedia: the Missing Manual) probably has a good claim for fair use (in the US, where the work is published) - they're certainly on a better moral ground than the hypothetical publisher who makes minor changes to a set of articles and sells printed copies without listing any authors. I found 15 cases of this ranking in the english data - 2.5% (whether they found linking "unacceptable" or merely "less acceptable" cannot be determined from the data given).
The only comparison you can make is between the option you are analysing and the status quo. In my opinion the status quo is closest to "Online: link; offline: list authors". Now Erik makes the claim that the current relicensing keeps with the spirit of the GFDL, which I guess means that he thinks the status quo and spirit of the GFDL is "link to any transparent copy that includes the same licensing and authorship information as the Wikipedia.org". But if that's the case, why doesn't the FSF just clarify the GFDL directly?
The direct, heads up comparison between "linking to the article" and "Online: link; offline: list authors" is 70-80 percent in favor of always allowing linking. Right in the grey area if this decision were the "no big deal" one of appointing an administrator to the English Wikipedia. Considering the nature of this particular proposed change, I think the right decision for the board to make is obvious - no consensus.
Some options may be out of the question due to local law. John
Erik Moeller skrev:
2009/3/7 Thomas Dalton thomas.dalton@gmail.com:
I'm curious, why did you include options that aren't actually available? No credit and credit to the community are clearly not in keeping with the license, so knowing who would accept them isn't particularly useful (although I'm not sure it hurts).
We tried to surface people's "true preference" for an attribution model. (While of course the provided options can't capture everything, the relatively low number of write-in options for additional attribution models suggests that respondents generally found their views represented somewhere in the continuum of given options.) People's true preferences should guide our thinking process, and if we clouded the available options with perceived or real constraints, we wouldn't be able to approximate the best feasible solution. It helps us to uncover both where people may be willing to compromise and where they may not be.
For example, if the survey had shown community credit to be highly desired and not controversial at all, that would be interesting: We could have an informed conversation about whether we should try to accommodate that model after all. As it is, it's the second most popular first option, but with 15.29% ranking it as their second-to-last option, it's also somewhat polarizing. A link to the article, on the other hand, is the first or second option for more than 60% of respondents, and the last or second-to-last option for only 3.47%.
Which[6] is interesting. Almost all responses were "no credit is needed" (453) or "full list of authors must always be copied" (322). 155 did not fill out Which[6] - presumably choices were left unranked when they were less acceptable than all the ranked choices, but this might be assuming too much. Of course, whether people prefer "no credit" or "full list of authors" isn't particularly relevant.
The rest of the figures in the pdf summary are much harder to interpret. No option gained a majority of first choices. A condorcet winner could probably be determined from the raw numbers, though.
Of course, as I've made widely known, I think this is a horrible thing to be voting on in the first place.
On Sat, Mar 7, 2009 at 5:25 PM, Anthony wikimail@inbox.org wrote:
A condorcet winner could probably be determined from the raw numbers, though.
I should add that the choices are mostly not mutually exclusive. If someone felt strongly that an article should contain *both* a link to the article *and* a list of all authors, and felt equally strongly that each should be there, there was no way to express that but to randomly place one choice above the other.
On Sat, Mar 7, 2009 at 2:25 PM, Anthony wikimail@inbox.org wrote: <snip>
A condorcet winner could probably be determined from the raw numbers, though.
<snip>
Condorcet Ranking (for the enwiki data):
1) Link to the article must be given. 2) Collective credit (e.g. Wikipedia community). 3) Link to the version history must be given. 4) For online use: link. For other uses: full list of authors. 5) Full list of authors must always be copied. 6) No credit is needed.
-Robert Rohde
On Sat, Mar 7, 2009 at 3:41 PM, Robert Rohde rarohde@gmail.com wrote:
On Sat, Mar 7, 2009 at 2:25 PM, Anthony wikimail@inbox.org wrote:
<snip> > A condorcet winner could probably be determined from the raw numbers, though. <snip>
Condorcet Ranking (for the enwiki data):
- Link to the article must be given.
- Collective credit (e.g. Wikipedia community).
- Link to the version history must be given.
- For online use: link. For other uses: full list of authors.
- Full list of authors must always be copied.
- No credit is needed.
-Robert Rohde
German data gave the same Condorcet ranking.
-Robert Rohde
On Sat, Mar 7, 2009 at 6:58 PM, Robert Rohde rarohde@gmail.com wrote:
On Sat, Mar 7, 2009 at 3:41 PM, Robert Rohde rarohde@gmail.com wrote:
On Sat, Mar 7, 2009 at 2:25 PM, Anthony wikimail@inbox.org wrote:
<snip> > A condorcet winner could probably be determined from the raw numbers,
though.
<snip>
Condorcet Ranking (for the enwiki data):
- Link to the article must be given.
- Collective credit (e.g. Wikipedia community).
- Link to the version history must be given.
- For online use: link. For other uses: full list of authors.
- Full list of authors must always be copied.
- No credit is needed.
-Robert Rohde
German data gave the same Condorcet ranking.
-Robert Rohde
Cool. So, personally, I'd be interested in the head's up ranking of 1) vs. 4). There would be 5 possibilities: 1 beats 4, 4 beats 1, 1 not ranked, 4 not ranked, 1 and 4 both not ranked.
The reason I wonder about this particular matchup is that I find 4 and 5 to be morally acceptable (but 4 beats 5), and 1 to be the obvious choice if moral considerations were to be ignored, so I wonder how strongly the preference of 1 over 4 is. Still doesn't answer the more important question, though, which is basically "how many contributors feel like they're being robbed if 1) gets implemented".
On Sat, Mar 7, 2009 at 4:53 PM, Anthony wikimail@inbox.org wrote:
On Sat, Mar 7, 2009 at 6:58 PM, Robert Rohde rarohde@gmail.com wrote:
On Sat, Mar 7, 2009 at 3:41 PM, Robert Rohde rarohde@gmail.com wrote:
On Sat, Mar 7, 2009 at 2:25 PM, Anthony wikimail@inbox.org wrote:
<snip> > A condorcet winner could probably be determined from the raw numbers,
though.
<snip>
Condorcet Ranking (for the enwiki data):
- Link to the article must be given.
- Collective credit (e.g. Wikipedia community).
- Link to the version history must be given.
- For online use: link. For other uses: full list of authors.
- Full list of authors must always be copied.
- No credit is needed.
-Robert Rohde
German data gave the same Condorcet ranking.
-Robert Rohde
Cool. So, personally, I'd be interested in the head's up ranking of 1) vs. 4). There would be 5 possibilities: 1 beats 4, 4 beats 1, 1 not ranked, 4 not ranked, 1 and 4 both not ranked.
<snip>
1 beats 4: 69% 4 beats 1: 21% 1 not ranked: 0.9% 4 not ranked: 4.4% neither ranked: 4.4%
-Robert Rohde
On Sat, Mar 7, 2009 at 8:24 PM, Robert Rohde rarohde@gmail.com wrote:
1 beats 4: 69% 4 beats 1: 21% 1 not ranked: 0.9% 4 not ranked: 4.4% neither ranked: 4.4%
-Robert Rohde
Pretty much the same for German. Between 70 and 80 percent, right at that gray area if this were an RfA. This should be fun to watch.
On Sat, Mar 7, 2009 at 4:53 PM, Anthony wikimail@inbox.org wrote:
On Sat, Mar 7, 2009 at 6:58 PM, Robert Rohde rarohde@gmail.com wrote:
On Sat, Mar 7, 2009 at 3:41 PM, Robert Rohde rarohde@gmail.com wrote:
On Sat, Mar 7, 2009 at 2:25 PM, Anthony wikimail@inbox.org wrote:
<snip> > A condorcet winner could probably be determined from the raw numbers,
though.
<snip>
Condorcet Ranking (for the enwiki data):
- Link to the article must be given.
- Collective credit (e.g. Wikipedia community).
- Link to the version history must be given.
- For online use: link. For other uses: full list of authors.
- Full list of authors must always be copied.
- No credit is needed.
-Robert Rohde
German data gave the same Condorcet ranking.
-Robert Rohde
Cool. So, personally, I'd be interested in the head's up ranking of 1) vs. 4). There would be 5 possibilities: 1 beats 4, 4 beats 1, 1 not ranked, 4 not ranked, 1 and 4 both not ranked.
The reason I wonder about this particular matchup is that I find 4 and 5 to be morally acceptable (but 4 beats 5), and 1 to be the obvious choice if moral considerations were to be ignored, so I wonder how strongly the preference of 1 over 4 is. Still doesn't answer the more important question, though, which is basically "how many contributors feel like they're being robbed if 1) gets implemented".
Incidentally, of the people who ranked both 4 and 5, 88% preferred 4 over 5 which was the highest margin on any pairwise matchup.
-Robert Rohde
On Sat, Mar 7, 2009 at 4:41 PM, Robert Rohde rarohde@gmail.com wrote:
On Sat, Mar 7, 2009 at 2:25 PM, Anthony wikimail@inbox.org wrote:
<snip> > A condorcet winner could probably be determined from the raw numbers, though. <snip>
Condorcet Ranking (for the enwiki data):
- Link to the article must be given.
- Collective credit (e.g. Wikipedia community).
- Link to the version history must be given.
- For online use: link. For other uses: full list of authors.
- Full list of authors must always be copied.
- No credit is needed.
-Robert Rohde
Nice work.
1, 2, 3, 4 or 5 at the re-user's option seems like a good compromise. I interpret 1 as meaning that the survey taker probably would be inclined to provide at least a link to the article if they were the one re-using the content. This means that most people understand that providing a link is an acceptable minimum level of attribution and would be inclined to do it whether or not the license forces them to. At the same time, many people seem OK with community credit, which can practically be shortened to just "Wikipedia, Article Title" or similar (in the case of the WikiBlame extension, just Wikipedia). This is useful for re-users on non-hypertext media where a protocol declaration and specification makes little sense, it being so easily and obviously constructed from just the simpler text attribution. Providing the specific revision number is an acceptable option when that revision number is appropriate and helpful. However, I see no reason to encourage sending people to old versions of article data when its not necessary. If someone wants to cite the full list of authors, it should be made readily available to them.
Ultimately it does not seem reasonable to force the printing of a URL on non-hypertext mediums.
This is exactly the same as 1. The reuser always has the option of providing more information, upto and including the text of every version in the history.
On Sat, Mar 7, 2009 at 9:13 PM, Brian Brian.Mingus@colorado.edu wrote:
On Sat, Mar 7, 2009 at 4:41 PM, Robert Rohde rarohde@gmail.com wrote:
On Sat, Mar 7, 2009 at 2:25 PM, Anthony wikimail@inbox.org wrote:
<snip> > A condorcet winner could probably be determined from the raw numbers, though. <snip>
Condorcet Ranking (for the enwiki data):
- Link to the article must be given.
- Collective credit (e.g. Wikipedia community).
- Link to the version history must be given.
- For online use: link. For other uses: full list of authors.
- Full list of authors must always be copied.
- No credit is needed.
-Robert Rohde
Nice work.
1, 2, 3, 4 or 5 at the re-user's option seems like a good compromise. I interpret 1 as meaning that the survey taker probably would be inclined to provide at least a link to the article if they were the one re-using the content. This means that most people understand that providing a link is an acceptable minimum level of attribution and would be inclined to do it whether or not the license forces them to. At the same time, many people seem OK with community credit, which can practically be shortened to just "Wikipedia, Article Title" or similar (in the case of the WikiBlame extension, just Wikipedia). This is useful for re-users on non-hypertext media where a protocol declaration and specification makes little sense, it being so easily and obviously constructed from just the simpler text attribution. Providing the specific revision number is an acceptable option when that revision number is appropriate and helpful. However, I see no reason to encourage sending people to old versions of article data when its not necessary. If someone wants to cite the full list of authors, it should be made readily available to them.
Ultimately it does not seem reasonable to force the printing of a URL on non-hypertext mediums.
foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
On Sat, Mar 7, 2009 at 9:13 PM, Brian Brian.Mingus@colorado.edu wrote:
1, 2, 3, 4 or 5 at the re-user's option seems like a good compromise.
Compromise? Between what two sides would that be a compromise?
Might as well make it 1, 2, 3, 4, 5 or 6 at the re-user's option. Now there's a compromise, eh? LOL.
Oh oh, here's a compromise. 1, 2, 3, 4, or 5 at the author's option. Or, how about 1, 2, 3, 4, *and* 5. LOL.
On Sat, Mar 7, 2009 at 9:34 PM, David Goodman dgoodmanny@gmail.com wrote:
This [1, 2, 3, 4 or 5 at the re-user's option] is exactly the same as 1.
I'd say it's much worse than 1. It's certainly not *exactly* the same.
Just to be clear, my argument does not allow for 6. Just because an alternate compromise is possible does not imply that my argument allows for such a compromise. As another person said, its effectively just the first.
On Sat, Mar 7, 2009 at 7:39 PM, Anthony wikimail@inbox.org wrote:
On Sat, Mar 7, 2009 at 9:13 PM, Brian Brian.Mingus@colorado.edu wrote:
1, 2, 3, 4 or 5 at the re-user's option seems like a good compromise.
Compromise? Between what two sides would that be a compromise?
Might as well make it 1, 2, 3, 4, 5 or 6 at the re-user's option. Now there's a compromise, eh? LOL.
Oh oh, here's a compromise. 1, 2, 3, 4, or 5 at the author's option. Or, how about 1, 2, 3, 4, *and* 5. LOL.
On Sat, Mar 7, 2009 at 9:34 PM, David Goodman dgoodmanny@gmail.com wrote:
This [1, 2, 3, 4 or 5 at the re-user's option] is exactly the same as 1.
I'd say it's much worse than 1. It's certainly not *exactly* the same. _______________________________________________ foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
2009/3/7 Brian Brian.Mingus@colorado.edu:
Ultimately it does not seem reasonable to force the printing of a URL on non-hypertext mediums.
I still believe we ought to avoid explicit distinction between media forms because I think these distinctions are inherently fragile. When you take the printed book and read it on the Kindle, or you have the DVD and a net connection, suddenly the link _can_ be clickable and meaningful. I think we can recommend a link to the article, and rely to the "reasonable to the medium or means" clause in CC-BY-SA to provide some flexibility for edge cases.
As per earlier discussions, the link clause should probably be phrased so that reference to any transparent copy that includes the same licensing and authorship information as the Wikipedia.org copy is sufficient credit, allowing for forks that retain authorship information while separating themselves from Wikipedia fully.
Erik Moeller wrote:
2009/3/7 Brian Brian.Mingus@colorado.edu:
Ultimately it does not seem reasonable to force the printing of a URL on non-hypertext mediums.
I still believe we ought to avoid explicit distinction between media forms because I think these distinctions are inherently fragile.
More importantly, I think they can become unjustly discriminatory. If we say we are making knowledge freely available, but then make the terms of that freedom more onerous offline than on, how free is that? Particularly when reliable internet access is already a sign of privilege and wealth. My sense is that people producing offline media don't complain much about being "forced" to provide URLs. Instead it's the people fully invested in the online world complaining when other media can't fully replicate the features of hypertext.
--Michael Snow
2009/3/8 Erik Moeller erik@wikimedia.org:
2009/3/7 Brian Brian.Mingus@colorado.edu:
Ultimately it does not seem reasonable to force the printing of a URL on non-hypertext mediums.
I still believe we ought to avoid explicit distinction between media forms because I think these distinctions are inherently fragile.
CC-BY-SA-3.0 clearly states credit must be reasonable to the medium or means. Distinction between media forms are none of the foundation's business since the issue is already taken care of by the license text.
One person told me that attribution of a single article and a bigger collection could be made different. That is, a single printed copy of an article could use a credit of "Wikipedia" and a mirror on a website could use a history link. We don't have to choose a "one scheme fits all" -solution.
john
Erik Moeller skrev:
The author attribution survey is now closed. We have 1017 complete responses. I've posted results of the attribution data in the following report:
http://wikimediafoundation.org/wiki/File:Attribution_Survey_Results.pdf
I've posted the raw data of the attribution survey here:
Respondents from English Wikipedia: http://wikimediafoundation.org/wiki/File:Attsurvey-en.ods Respondents from German Wikipedia: http://wikimediafoundation.org/wiki/File:Attsurvey-de.ods Respondents from miscellaneous languages and projects: http://wikimediafoundation.org/wiki/File:Attsurvey-misc.ods
(The survey was linked via the WikimediaNotifier bot, so we got quite a bit of nicely dispersed traffic.)
As the report shows, and as I indicated in my prior e-mail, there is wide support for simple attribution models, and fairly strong and visible opposition to full author attribution (as well as complete absence of any attribution). Full author attribution is the second least popular option, at 32.82%. Many comments pointed out the tension between free content and attribution, such as:
- "While the whole point of Wikipedia is to provide access to
information freely and easily, a balance must be struck between recognising authors' contributions and the constraints on utilising the information." (User's preferred attribution model is link to the article.)
- "Giving credit to all authors is ridiculous! I think the 'Wikipedia
Community' is sufficient credit, this project is not about personal gratification, its about community collaboration."
- "Full list of authors is terribly impractical."
- "Including the full list of authors on a 'NOT online' resource would
be a waste of resources, i.e. paper and ink, most of the time. But even for online use, who would read the version history? On the other hand, a link can't do much harm..."
- "Establishing which editors to credit would cause enormous disagreement"
- "Although requiring credit may sound noncontroversial, it actually
is a pretty big can of worms in contexts of (a) editing wikipedia-sourced content into rather different things (for example, the way that some wikipedia articles grew out of 1911 Britannica articles), (b) what if the wikimedia foundation has some kind of meltdown and it is necessary to fork the project. Therefore my recommendation is to not think in terms of 'requirements' but suggested practices."
Some users commented on the fact that Wikipedia is primarily written by people under pseudonyms, and that being suddenly visibly attributed would actually come as a surprise:
- "If any version of credit-sharing citing editors is made policy, all
editors should be given notice and allowed to change their monikers to their choice. In my case, I choose a moniker I liked when I thought the community would remain anonymous forever. If my contributions went into print or were used similarly I would like to use my actual name."
Community credit proved a quite popular option, second only to a direct link to the article. Many people viewed it as a simple method to credit their contribution both online and offline. (At least one user suggested linking to detailed histories online, and crediting the community collectively offline.)
A few users felt very strongly about always giving author credit. The strongest example I found:
"I won't accept nothing less than what I chosed above, and I'm ready to leave my sysop status and other wmf-related roles if WMF will underestimate the meaning of GFDL to our projects. GFDL is what we would have chosen if asked 8 years ago, and is what we will stand up for."
Some users also pointed out that our options were constrained by the requirements set forth in the GFDL.
I'd love to see deeper analysis of the survey. I want to restate my original intent in running it: it's intended to be a feeler survey, to get a rough impression of what attribution models are widely considered acceptable by contributors to our projects, and which ones aren't. It served this purpose, and I have no intent in running additional surveys; we're on an aggressive timeline and have to move forward. It's also not intended to dictate a solution.
My preliminary conclusion is that a simple, manageable attribution model, while causing some short-term disruption, will widely be considered not only acceptable, but preferable to complex attribution models, in support of our mission to disseminate free information. That being said, we probably still have to find a compromise, as well as language that appropriate deals with single-author multimedia contributions. I imagine that if we a) have a more prominent "list of authors / list of people who contributed to this revision" credit link on article pages; b) require that a link must be given, and that the preferred linking format is to the revision that is being copied, c) explicitly state in our attribution terms that for images, sounds and videos that aren't the result of extensive collaboration, credit must be given to the creator, we're covering most cases.
We then still have to resolve the issue of giving credit for content imported into our projects consistently, which is a bit of a can of worms. (We might want to set some limitations on what kinds of content we import, to prevent "attribution pollution".) But it's secondary to the main issue of a consistent attribution model within our projects.
A model like the above is consistent with CC-BY-SA. There is a question as to whether it can be reconciled with our current practices. I believe it can, and I also think we can find mitigation strategies for contributors who vehemently disagree. I'll work on a revision to the currently proposed language, and will post that next week, alongside some further thoughts.
In terms of our timeline, I don't believe we can wrap things up prior to the Board meeting in April, but I think we can still hit a timeline to make a migration decision by mid-to-late April. SPI has committed to help administer the vote as an independent third party. What still needs to be done:
- We need to form a little workgroup/committee to help with the usual
process of tallying the votes;
- We need to translate all relevant text (including the vote
announcements), once it's final, into as many languages as possible;
- We need to implement a modified Special:Boardvote so it can be used
for this decision.
- We also want to allow sufficient time for the actual
decision-making, ideally 3-4 weeks.
We have a big all-staff meeting and an all-day tech meeting next week, which will hamper us a bit in moving this forward aggressively, but I'll see if I can move things along a bit before then. If someone wants to create draft pages for any of the above (workgroup, announcement, etc.), I'd be very grateful :-)
More soon, Erik
2009/3/8 John at Darkstar vacuum@jeb.no:
One person told me that attribution of a single article and a bigger collection could be made different. That is, a single printed copy of an article could use a credit of "Wikipedia" and a mirror on a website could use a history link. We don't have to choose a "one scheme fits all" -solution.
Crediting "Wikipedia" would never be acceptable under CC-by-SA, since existing contributions weren't made under a terms of service that required permission be granted for such attribution. However, your example not working doesn't mean your point is incorrect. We're trying to determine what form of attribution is reasonable. It stands to reason that what is reasonable will depend on the context, and amount of content being used could well be a factor. There is an argument for keeping our recommended forms of attribution (and that's all we're talking about here - we can't demand a certain form of attribution, just recommend one) as simple as possible, though.
On Sun, Mar 8, 2009 at 11:04 AM, Thomas Dalton thomas.dalton@gmail.comwrote:
Crediting "Wikipedia" would never be acceptable under CC-by-SA, since existing contributions weren't made under a terms of service that required permission be granted for such attribution.
True, but would the Creative Commons lawyers agree with you? Typing "Wikipedia Alabama" into my browser brings up the same page as typing " http://en.wikipedia.org/wiki/Alabama".
2009/3/8 Anthony wikimail@inbox.org:
On Sun, Mar 8, 2009 at 11:04 AM, Thomas Dalton thomas.dalton@gmail.comwrote:
Crediting "Wikipedia" would never be acceptable under CC-by-SA, since existing contributions weren't made under a terms of service that required permission be granted for such attribution.
True, but would the Creative Commons lawyers agree with you? Typing "Wikipedia Alabama" into my browser brings up the same page as typing " http://en.wikipedia.org/wiki/Alabama".
Yes, but there is even less guarantee that that will always work than there is that the URL will always work. And it relies on you knowing that it should work and trying it.
I'm not sure if I would like to credit "Wikipedia" anyhow, Wikipedia is not the author even if tradition says you can give attribution to an encyclopedia in some countries. I think GFDL is better on this, even if the current practice on Wikipedia is crappy on attribution. The main authors of an article should be identified anyhow, and simpler schemes should only be used when other solutions are impractical. John
Thomas Dalton skrev:
2009/3/8 Anthony wikimail@inbox.org:
On Sun, Mar 8, 2009 at 11:04 AM, Thomas Dalton thomas.dalton@gmail.comwrote:
Crediting "Wikipedia" would never be acceptable under CC-by-SA, since existing contributions weren't made under a terms of service that required permission be granted for such attribution.
True, but would the Creative Commons lawyers agree with you? Typing "Wikipedia Alabama" into my browser brings up the same page as typing " http://en.wikipedia.org/wiki/Alabama".
Yes, but there is even less guarantee that that will always work than there is that the URL will always work. And it relies on you knowing that it should work and trying it.
foundation-l mailing list foundation-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
Erik Moeller wrote:
My preliminary conclusion is that a simple, manageable attribution model, while causing some short-term disruption, will widely be considered not only acceptable, but preferable to complex attribution models, in support of our mission to disseminate free information. That being said, we probably still have to find a compromise, as well as language that appropriate deals with single-author multimedia contributions. I imagine that if we a) have a more prominent "list of authors / list of people who contributed to this revision" credit link on article pages; b) require that a link must be given, and that the preferred linking format is to the revision that is being copied, c) explicitly state in our attribution terms that for images, sounds and videos that aren't the result of extensive collaboration, credit must be given to the creator, we're covering most cases.
Whatever the model chosen, enforceability still needs to be taken into account. Copyright laws have worked for the last three centuries because the rights owners were willing to accept the responsibility of enforcing those rights. In recent years there has been a greater shift to treating of these in quasi-criminal legislation, thus transferring the responsibility for enforcement to governments. This creates a separation between the rights and the responsibilities.
If we grant individual authors the right to be directly attributed, who will enforce that. If that individual clearly contributed only one identifiable sentence in a long article (assuming that that is at all determinable), is it reasonable to expect that he will undertake judicial proceedings to enforce his economic rights to that one sentence? Rightly or wrongly, we certainly have no evidence that the Wikimedia Foundation will ever take legal action to protect the collective legal rights of all the contributors to an article. Who does that leave to enforce rights? Governments? That too seems unrealistic, even where copyrights have been made a part of the criminal law.
I am a frequent buyer of books from eBay. Occasionally when I'm shopping there I encounter a book description that uses material from Wikipedia, and credits it as such. Some even include the red links to non-existent articles. Can we expect anything more there? Are we going to convince eBay to cancel an auction offering because the seller failed to include a full list of the authors to the material that he used, or failed to sort out the authors of the part of the article that he used? If so, who will be the "we" who takes the responsibility to make sure that this gets done?
We waste a lot of electrons worrying about what might be proper attribution, or the fine points of distinction between GFDL and CC, with very little, if any, litigation to back any interpretation of these licences. Sometimes I wonder if all of this activity is anything more substantial than an exercise in stacking intellectual Jenga blocks.
The principles of free access are just fine, and I am left to wonder whether a statement of such principles would be just as or more effective as an elaborate licensing scheme.
Ec
wikimedia-l@lists.wikimedia.org