Hi,
there's been some movement forward on the Single User Login (SUL) issue. I ask the Board to review this mail carefully as this has significant long- term implications and we need Board input to go ahead. I also ask other developers to correct me if I misrepresent anything.
There are currently three competing strategies. Before I describe these strategies, let me point out that one important consideration for any system is scalability. That is, single login will be used on all existing and future Wikimedia projects, and potentially even on non-Wikimedia sites which we allow to participate in our system.
The three strategies are:
1) GLOBAL NAMESPACE, IMMEDIATE CONFLICT RESOLUTION
We try to move towards a single global user namespace for all Wikimedia wikis. If a name is already taken in the global namespace, you have to find one which isn't.
For the migration, any names which clearly belong to the same user are combined into one. If passwords and email addresses are different, the user can manually link together any accounts which belong to him by providing the passwords.
For cases of true name conflicts between the existing wikis, there is a resolution phase, where factors like seniority, use on multiple wikis vs. a single one, etc., are weighed in - the "loser" has to choose a new account name.
After the manual resolution phase, any remaining accounts are converted to the new system automatically by making them unique, e.g. by adding a number to the username. The transition is now complete. The old system no longer exists.
2) GLOBAL NAMESPACE COEXISTING WITH LOCAL ONES, DEFERRED CONFLICT RESOLUTION
As in 1), we migrate all existing accounts to the new global namespace automatically if possible. New accounts are created in the global namespace.
Where there are people sharing the same name, the accounts will not be migrated to the global namespace but will stay in the local ones, which will continue to coexist. These people can keep using their local IDs, or create a new, different global identity.
The idea here is that resolving name conflicts is so complicated that we simply defer the issue for now.
3) GLOBAL COMPUTER-READABLE ID, LOCAL HUMAN-READABLE NAMES
Every user has a global, numeric ID which is unique. But for each wiki, they can have a different username. As in strategy 1), any clearly identical accounts will be linked to a single GUID automatically, others can be hooked up by providing the passwords.
Naming conflicts are not an issue in this system. Let's say I am Eloquence on en: and de:, and there is another Eloquence on fr:. I get the global ID 1233, the fr: user gets the ID 28387. When I go to fr: and try to edit a page, I get a prompt:
The username you have chosen is already in use on this wiki. Please specify a new name: ____________________________________
If this is your account and it has not been linked to your global ID yet, please provide the password: ___________________________________
If I go to another wiki with no user named Eloquence on it, however, the local username "Eloquence" will be reserved for me the moment I edit it or set my local prefs. This is because the system knows that "Eloquence" is my preferred username, and will automatically try to create it for me when I need it. I can change the name later, if I want, and use different names on different wikis.
- - - - -
The differences:
1) is very complex, and we may not find someone willing to deal with the name conflict resolution issue and take the blame from annoyed users at the same time. Naming conflicts will always be an issue in this scheme, as e.g. all common first names will be taken, and any small wiki hooking up with our SUL system would feel this impact. People can mutate these usernames relatively easily to make them unique - Erik333 - and the system can offer such mutations, but it's still a bit annoying.
2) is easiest to implement by avoiding the conflict issue. Brion has expressed interest in coding this. It will annoy some people who can choose between using their local name and keeping their attributions, or a global one, and losing their attributions, at least until things like linking up local accounts to a global one are implemented. It leads to some ugliness in the system because we are in an "in-between" state for now.
3) does not have the naming conflict problem. Both Jamesday and Kate have expressed interest in implementing it. Jamesday also wants to write a more detailed proposal on Meta about it. For the user, it is fairly straight- forward -- existing accounts can be hooked up easily to the single login, new names are picked only when necessary. It is somewhat vulnerable to trolling, as someone could e.g. register the name Eloquence on a wiki where I am not active, and use it for nefarious purposes. The system could however make it fairly easy to find out that this is not the same person (on the user page: "This user is active on the following wikis: .. under these names: ..").
In any case, we want username changes to be a cheap operation, so that anti-trolling policies should be easily enforcable.
The question for the Board is: Given that we have at least two, and possibly three, implementations where there are volunteers, which one should we choose? To me, this seems to be a matter that could be decided by the Board; others have said that a poll will be necessary first. Either way is fine with me - what does the Board say?
Regards,
Erik
Not that I'm a Board member, but, just to give my input, Option 3 sounds like the easiest and fairest to me. It would be easy to link accounts under this proposed system, even those with different names. I envision some sort of "global user page" that can be accessed through any local Wiki user page with the user's accounts on all projects listed, along with any pertinent statistics. I think the trolling issue can be addressed by allowing some sort of "eminent domain"; that is, users with significant contributions on a project or projects can take over an account on a project if it is inactive/has few contributions; the original account would be moved to an alternate user name. Those are my two U.S. cents.
--Slowking Man
On 12 Nov 2004 04:27:00 +0100, Erik Moeller erik_moeller@gmx.de wrote:
Hi,
there's been some movement forward on the Single User Login (SUL) issue. I ask the Board to review this mail carefully as this has significant long- term implications and we need Board input to go ahead. I also ask other developers to correct me if I misrepresent anything.
*snip*
The question for the Board is: Given that we have at least two, and possibly three, implementations where there are volunteers, which one should we choose? To me, this seems to be a matter that could be decided by the Board; others have said that a poll will be necessary first. Either way is fine with me - what does the Board say?
Regards,
Erik _______________________________________________ foundation-l mailing list foundation-l@wikimedia.org http://mail.wikipedia.org/mailman/listinfo/foundation-l
The question for the Board is: Given that we have at least two, and possibly three, implementations where there are volunteers, which one should we choose? To me, this seems to be a matter that could be decided by the Board; others have said that a poll will be necessary first. Either way is fine with me - what does the Board say?
I don't feel this ought to be a board decision at the moment. I suggest a non-binding poll of the community take place first. This can be used to inform the developers and also to help to adjust the current proposals for single-login. I'd rather the developer committee came to a consensus about this issue, based on the results of the poll, since it is them who have to create the single login feature. There is little point in the board saying make it all magically work for an option that is disliked by the community or not likely to be coded.
Angela.
Well i just chased down my first ever vandal (221.219.60.170 on en.wikibooks) and i think that which ever option is taken, its going to make tracking down cross wikivandals far easier. also if we did that, we could presumably have a cross-wiki recent changes page which might help protect the smaller wikis by giving them the protection from the RC patrol from the big wikiprojects. I personaly support version 1 or maybe 3. Anyway, best of luck to the developers. Is it possible to create a list now of how many clashes there would be with option 1? so that whomever(s) makes the decision are informed.
paz y amor, rjs [[User:the bellman]]
On Fri, 12 Nov 2004 08:54:28 +0000, Angela beesley@gmail.com wrote:
The question for the Board is: Given that we have at least two, and possibly three, implementations where there are volunteers, which one should we choose? To me, this seems to be a matter that could be decided by the Board; others have said that a poll will be necessary first. Either way is fine with me - what does the Board say?
I don't feel this ought to be a board decision at the moment. I suggest a non-binding poll of the community take place first. This can be used to inform the developers and also to help to adjust the current proposals for single-login. I'd rather the developer committee came to a consensus about this issue, based on the results of the poll, since it is them who have to create the single login feature. There is little point in the board saying make it all magically work for an option that is disliked by the community or not likely to be coded.
Angela.
foundation-l mailing list foundation-l@wikimedia.org http://mail.wikipedia.org/mailman/listinfo/foundation-l
Angela a écrit:
The question for the Board is: Given that we have at least two, and possibly three, implementations where there are volunteers, which one should we choose? To me, this seems to be a matter that could be decided by the Board; others have said that a poll will be necessary first. Either way is fine with me - what does the Board say?
I don't feel this ought to be a board decision at the moment. I suggest a non-binding poll of the community take place first. This can be used to inform the developers and also to help to adjust the current proposals for single-login. I'd rather the developer committee came to a consensus about this issue, based on the results of the poll, since it is them who have to create the single login feature. There is little point in the board saying make it all magically work for an option that is disliked by the community or not likely to be coded.
Angela.
I agree with Angela on this.
Anthere
As a side comment, I do not use same password on all my wiki accounts. I probably still use an old one on a few projects I recently change the french one due to a security breach in Paris meeting I use different password for meta and wikimediafoundation because of the higher risks involved to my password being stolen there
That makes a least 4 different passwords.
Erik Moeller wrote: <snip>
The question for the Board is: Given that we have at least two, and possibly three, implementations where there are volunteers, which one should we choose? To me, this seems to be a matter that could be decided by the Board; others have said that a poll will be necessary first. Either way is fine with me - what does the Board say?
Regards,
Erik
Hello,
Maybe we should analyse the usernames and try to get data about conflicting name. Maybe there is only a handful of conflict ?
On Fri, 12 Nov 2004 13:36:42 +0100, Ashar Voultoiz hashar@altern.org wrote:
Erik Moeller wrote:
<snip> > The question for the Board is: Given that we have at least two, and > possibly three, implementations where there are volunteers, which one > should we choose? To me, this seems to be a matter that could be decided > by the Board; others have said that a poll will be necessary first. Either > way is fine with me - what does the Board say? > > Regards, > > Erik
Hello,
Maybe we should analyse the usernames and try to get data about conflicting name. Maybe there is only a handful of conflict ?
Kate ran a query on just de and en and there were several thousand conflicts (about 3000 or 4000 whose passwords and/or emails did not match).
On Fri, 12 Nov 2004 07:32:09 -0600, Dori slowpoke@gmail.com wrote:
Kate ran a query on just de and en and there were several thousand conflicts (about 3000 or 4000 whose passwords and/or emails did not match).
How many of those are 'real' though? If you look at passwords and/or emails not matching, it means that those where one matches are also counted as conflicts. I think we should not count matching emails with different passwords as conflicts, and different emails with the same password only if the password is the empty string. In fact, I alone could easily count as over 100 conflicts, with over 50 logins under the same name and at least 3 different email addresses.
Andre Engels
On Fri, 12 Nov 2004 16:34:34 +0100, Andre Engels andreengels@gmail.com wrote:
Kate ran a query on just de and en and there were several thousand conflicts (about 3000 or 4000 whose passwords and/or emails did not match).
How many of those are 'real' though?
'Real' in what sense? In the sense of 2 active users actively using the same name on different wikis, there is no sure way to tell until we start telling people to migrate; but if there are as many as 4000 *potential* clashes just between these 2 wikis, it's pretty certain even that's going to be no small number once you multiply up by the several-hundred wikis we now have running...
In fact, I alone could easily count as over 100 conflicts, with over 50 logins under the same name and at least 3 different email addresses.
And from a technical point of view, those are very real conflicts: ones which software alone cannot merge. Sure, we could have a publicity campaign immediately before/after the changeover (depending on the exact details of transition) to persuade people like yourself to manually remove such conflicts, but that is part of the process of transition nonetheless, and the scale of it has an impact on how we implement the transition.
--- Erik Moeller erik_moeller@gmx.de wrote:
Hi,
there's been some movement forward on the Single User Login (SUL) issue.
I would prefer #3.
I think this is the solution with the least long term issues.
Only problem: Do you need your userid number to log in?
===== Chris Mahan 818.943.1850 cell chris_mahan@yahoo.com chris.mahan@gmail.com http://www.christophermahan.com/
__________________________________ Do you Yahoo!? Check out the new Yahoo! Front Page. www.yahoo.com
On Fri, 12 Nov 2004 07:18:09 -0800 (PST), Christopher Mahan chris_mahan@yahoo.com wrote:
I would prefer #3.
I think this is the solution with the least long term issues.
Only problem: Do you need your userid number to log in?
I think no. You login on some wiki, and use your login name and password from that specific wiki. Then the system connects that login name to a userid, and that userid would be stored in or connected to your cookie. At least that's how I envision things.
Andre Engels
--- Andre Engels andreengels@gmail.com wrote:
On Fri, 12 Nov 2004 07:18:09 -0800 (PST), Christopher Mahan chris_mahan@yahoo.com wrote:
I would prefer #3.
I think this is the solution with the least long term issues.
Only problem: Do you need your userid number to log in?
I think no. You login on some wiki, and use your login name and password from that specific wiki. Then the system connects that login name to a userid, and that userid would be stored in or connected to your cookie. At least that's how I envision things.
Andre Engels
Scenario:
Person A is Joe on EN Person B is Joe on FR
Person A has ID 123 Person B has ID 456
Person A goes to EN, logs in as Joe, and gets ID 123, then goes to FR, and everythig is great.
While in FR, session times out. Person A, now in FR, tries to login again, as Joe. Oops, not possible, user Joe has a different password (because Joe on FR is really person B)
So you would have to go to the wiki that you originally registered your username with in order to log in.
===== Chris Mahan 818.943.1850 cell chris_mahan@yahoo.com chris.mahan@gmail.com http://www.christophermahan.com/
__________________________________ Do you Yahoo!? Check out the new Yahoo! Front Page. www.yahoo.com
Option 3 looks best to me. The trolling issue would be no worse than it is now. One way to *improve* that potential issue would be to grandfather-in currently conflicting names as proposed in option 3 but lock user names from now on.
-- mav
__________________________________ Do you Yahoo!? Check out the new Yahoo! Front Page. www.yahoo.com
Daniel Mayer wrote:
Option 3 looks best to me. The trolling issue would be no worse than it is now. One way to *improve* that potential issue would be to grandfather-in currently conflicting names as proposed in option 3 but lock user names from now on.
Yes. What I would say is that we should go with option 3, and that we should prevent any *future* conflicts by making it simple for people who do not currently have a name conflict to reserve their username globally if they like (this should be the default for new accounts).
For people with name conflicts, this can be a valuable inducement for them to eliminate the conflict - as long as you're conflicted, you have no guarantee on the username in a new wiki.
And finally, I would agree with Angela that this should not be a board decision at this point, but rather a very widely publicized community poll.
--Jimbo
Erik Moeller wrote: [cut]
The three strategies are:
[cut]
Those ideas are interesting but is there room for a other option?
I have a idea for a strategy that has some special benefits and of course some drawbacks. I would like to know of my idea is feasible and desirable.
(I think i have expressed this idea before but i do not remember the responds)
The idea is to create the concept of a home-wiki. The user name is expanded whit the language and name of the project.
My user name of the dutch Wikipedia is "walter". The expanded user name would be "walter@nl.wikipedia.org". My accounts on other wikimedia wikis can be renamed to the full user name from my home-wiki.
When I edit on the dutch wikipedia (my home-wiki) is am known as user "Walter". When I go to a other Wikipedia I can login whit my full user name. When I edit on the a other Wikipedia I am known whit the user name "walter@nl". When I go to a other non-Wikipedia Wikimedia project, like WikiQuote I am know as user "walter@nl.wikipedia.org"
This the general idea.
The good part of this idea is that there are no naming conflicts. No user will have to pick a new user name.
Also it is more fair. When you are a old user you can have a nice user name. Like when you would have the emailadress tom@hotmail.com
But there can no other Tom besides you on all the wikis whit the central login. The first user whit the name "Tom" on a new project must take a user name like "Tom1978_" because the other are long taken.
Whit my idea new user on every project has a fair change to have a nice name.
The bad part is the you have to pick a "home wiki". You have to make a choice between all projects and languages and say, this is my wiki. On all the other Wikis it will be clear that your a "foreigner" on that wiki and form where you are.
I have not problem whit this and I do not think that most users would have problems whit it. But some may have.
For those whit real problems whit it the could be provided whit a neutral user name like "username@wikimedia.org"
The first of the two conferences went quite well. I only ran into a couple of problems and these could perhaps be solved for the next conference.
It was held at Wentworth-by-the-Sea in Newcastle NH, right next to Rye and Hampton Beach... the place was gorgeous! And the facilities were awesome, the AV folks there were very helpful in getting the equipment all connected and very accomodating.
However the costs were pricey, an internet live connect went for $150. A live AV support person was $500 for the day... I didn't need an AV person but even so, they were helpful anyway... no charge for getting the connectors and making sure I was ok to start the presentation, they were very helpful indeed and very friendly!
All in all I presented on Friday and Saturday and the headcount totals were over 100 educators, school admins and staff from Northern New England area.
The presentation on Wikipedia was well received... the major objection, or should I say concern, is the question that "how can Wikipedia content be accurate without 'experts' writing the articles?"
My response is that the articles could be checked against other online ecyclopedia content and thus one could be reassured of the quality of content. Also that there are safeguards like 'peer editing' and 'npov' which help make the articles both accurate and non-biased, or as much as a npov objective can guarantee.
If there are other good points to counteract this objection I'd be interested in putting this into the next presentation... although for this next one at McAuliffe Center Tech Conference I'll have a fraction of the time I had at the NHAWLT presentation.
One of the problems I ran into was that I lost 'mousage', I lost the mouse pointer, when I projected the laptop screen. I used two different projectors and also tried an attached USB mouse instead of the glidepoint.
But the instance of MediaWiki running on the Linspire laptop was flawless of itself... a really good example of how powerful and yet easy to use and learn to edit the wiki engine is...
OpenOffice was very handy as a presentation tool. I used the suggested presentation from French and edited it and thinned it down for the time I had to work with... I also used the screen shots of the different language Wikipedias (that was done around February)... And I incorporated the Alexa charts to compare with Britannica, Groliers, Ecyclopedia.com and refer.com so as to give some idea of Wikipedia's growth and expansion.
I also briefly explained some of the other sister projects and highlighted why the wiki software lends itself so well to collaborative projects...
The next conference is in December.
With thanks to all who helped get me get MediaWiki running on Linspire and also those that gave me good suggestions on what to say during the presentation!
Best regards, Jay Bowks [[w:en:User:ILVI]]
On Wed, 27 Oct 2004 18:42:54 +0200, Ashar Voultoiz hashar@altern.org wrote:
ilooy wrote:
Hi fellow Wikipedians,
I'll be doing a presentation on Wikipedia for the McAuliffe Conference and the NHAWLT Teachers' Conference in the coming weeks and was wondering if anyone else has done a presentation to a large group... what features would be good to highlight... and what seems to work well with a group that has not heard much about the project yet.
If you have some ideas as to what might be good points to bring out please let me know. I appreciate any insights you may offer on this subject.
with sincere regards, Jay B. [[w:en:User:ILVI]]
new names are picked only when necessary. It is somewhat vulnerable to trolling, as someone could e.g. register the name Eloquence on a wiki where I am not active, and use it for nefarious purposes. The system could however make it fairly easy to find out that this is not the same person (on the user page: "This user is active on the following wikis: .. under these names: ..").
Isn't it possible to allow certain users (maybe anyone with more than 50 edits) to choose to automatically register his/her username on all available wikis?
As for option 1, I strongly oppose any system which would remove or change the attribution of users edits without their permission. This would almost surely not affect me personally (I doubt there are other users named "Anthony DiPierro" and even if there are I would assume I'd have "seniority"), but I consider it unethical to do it to anyone for any reason.
On a slightly related note, is there any intention to enable the "Real Name" feature on the wiki? Again it's not such a big deal for me personally, as I already use my real name as my username on most wikis where I contribute, but I can see how "Bob Jones #788888" might want his real name to appear somewhere in the history information.
Regards,
Erik
Anthony
wikimedia-l@lists.wikimedia.org