"Not necessarily. There is some complex mathematics involved[....] Basically a bucket of points is poured into the network by the roots, and it flows according to certification level. There are also sinks, and the whole shebang results in the cert list.
It could be that it currently works that way, but it is certainly not a rule appearing somewhere in the flow algorithm."
- Cees de Groot http://macos.tuwien.ac.at:9009/409183724.asHtml
Imagine someone who tries to join the Squeak community but has too low of a reputation to be allowed by the system. Aren't they going to be a little mystefied about the reason? Won't they wonder whether they are truly not Squeaky enough, versus not persistent enough in pestering people? Similarly, when someone gets added, what will existing members think? Will they feel confident that even one person in the community took a close look? Won't they wonder, just a little, if the new member merely pestered everyone until someone bumped them in?
A traditional membership system would involve at least one person in the community taking a serious look at each applicant. Frequently, there would additionally be a committee that takes a moderate look once the initial person is satisfied. This approach has advantages in both directions. The existing community knows that any new person has been reviewed thoroughly by *someone*, as opposed to getting a pile of reputation bumps by a bunch of people not really paying attention. Further, the applicant gets someone to interact with closely, so that they know what they are lacking if there is a problem.
This latter approach is used extraordinarily widely in existing communities. As just a few quick examples, it is used for:
- getting a Ph.D. - becoming a Methodist minister - earning a black belt - becoming a Debian developer
Anyway, none of this answers the question of who is in the *initial* member set. Using a reputation system still requires picking an initial set of members. Who do we pick?
-Lex
I dont expect the SqP mechanism to add human contact to the process - it is merely a way of expressing the personal contact and evaluation that is already happening. As to initial membership, I'm proposing we use the current status as the starting point.
This is my proposal. It has the following advantages in my opinion - it is already implemented, it is already much more transparent than most systems (you know who's for you), and its implementation is an algorithm, so that its properties can be explored effectively for bugs, and transparency can be increased by mere implementation work.
Lex, if you think we should use some other system based on the list of criteria discussed, please explain how you would implement it in detail. It would also be nice if you can give some sort of glimpse into who that system lets in, like we can get by peeking at the SqP site.
Anyway, here is my initial requirements list for a voting system. Functionality: 1. Allows a master to propose a topic on the website. A mail is sent to some every Squeaker that hasn't opted out. The topic proposer can also edit or remove the proposal. 2. Allows any Journeyman to propose an alternative course of action for this topic, and edit or remove it. An "ignore this topic" course of action is there by default. 3. Topic proposer gets to set a vote date. At this point another mail is sent to voting members that didn't opt out. 4. Journeymen can vote until said date, and gets confirmation mail whenever his vote changes. 5. Numeric results of Condorcet votes are published on the website as per Debian standards. Non-functional requirements: A. Easy backup. B. Topics, proposals, and votes are in files readable by text editor. Initially, we favor debuggability, and therefore reject privacy requirements.
Lex Spoon wrote:
"Not necessarily. There is some complex mathematics involved[....] Basically a bucket of points is poured into the network by the roots, and it flows according to certification level. There are also sinks, and the whole shebang results in the cert list.
It could be that it currently works that way, but it is certainly not a rule appearing somewhere in the flow algorithm."
- Cees de Groot
http://macos.tuwien.ac.at:9009/409183724.asHtml
Imagine someone who tries to join the Squeak community but has too low of a reputation to be allowed by the system. Aren't they going to be a little mystefied about the reason? Won't they wonder whether they are truly not Squeaky enough, versus not persistent enough in pestering people? Similarly, when someone gets added, what will existing members think? Will they feel confident that even one person in the community took a close look? Won't they wonder, just a little, if the new member merely pestered everyone until someone bumped them in?
A traditional membership system would involve at least one person in the community taking a serious look at each applicant. Frequently, there would additionally be a committee that takes a moderate look once the initial person is satisfied. This approach has advantages in both directions. The existing community knows that any new person has been reviewed thoroughly by *someone*, as opposed to getting a pile of reputation bumps by a bunch of people not really paying attention. Further, the applicant gets someone to interact with closely, so that they know what they are lacking if there is a problem.
This latter approach is used extraordinarily widely in existing communities. As just a few quick examples, it is used for:
- getting a Ph.D.
- becoming a Methodist minister
- earning a black belt
- becoming a Debian developer
Anyway, none of this answers the question of who is in the *initial* member set. Using a reputation system still requires picking an initial set of members. Who do we pick?
-Lex _______________________________________________ Elections mailing list Elections@lists.squeakfoundation.org http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/elections
It's barely possible that the Squeak People designations look strange to me simply because all of the people I am thinking of really have left Squeak. But let me take one more stab at showing you how strange it looks to me. Here are the people I posted before, this time with annotations. None of them will be allowed to vote if we "just use SqueakPeople". Don't you guys agree that leaving these people out is rather bizare?
Mark Guzdial, who wrote PWS and PWS/Swiki, who taught a Squeak-based university class to thousands of people over the years, who sponsored several Squeak-based Ph.D. projects, who authored a text book based on Squeak, and who co-edited the nu blue book. Bolot Karimbaev, author of Comanche. Alan Kay. KK Lamberty, a Ph.D. student who did the DigiQuilt project in Squeak. John Maloney, who among many other things brought Morphic to Squeak, gave Squeak a sound infrastructure, and is now working on the Scratch project. Jeff Pearce, who brought "Alice" to Squeak. Jeff Rick, who put many hours into building and refining ComSwiki, and who has put in years of scholarly work on wikis. Kim Rose, who has been with Squeak Central since the beginning, is a major player at ViewPoints, who co-edited the nublue book with Mark Guzdial. Jim Rowan, a Ph.D. student who did multiple projects in Squeak. Nathanael Shaerli, who gave Squeak the Genie gesture-recognition system and who worked out the traits system of restricted multiple inheritance.
Instead of rejecting all of these guys, here is a proposal based on the previously posted list of Squeak-based criteria.
For the first round of membership, let people apply on a Wiki page and post their reason for being considered a member. Objections and requests for more information can be posted in Wiki style. After a month or so, I expect that a consensus will emerge on practically all applicants (and, in fact, likely 100% of them will be included), and that will be the initial membership.
This seems easier, on the whole, than trying to invigorate the entire community to actually take part on Squeak People.
For later rounds of membership, I propose the same kind of thing, except that instead of having public review, we have one person in the group review them carefully and then post their findings publically somewhere. If no one objects after a certain period of time (one month?), then a person can then be included in the Squeak club with full privilages.
Again, this seems easier on the whole than trying to get people to maintain their Squeak People ratings.
For referenda, limiting who can propose them means that power gets rather centralized. Allowing resolutions to be proposed by anyone is a helpful check against the leadership getting carried away against the wishes of the group. Instead of defining an inner circle and using that to limit referenda proposals, how about requiring proposed referenda to be seconded? There are easy ways to avoid trivial referenda if that becomes a problem, but once power is centralized it is really hard to move in the opposite direction.
Daniel's non-functional requirements look excellent to me. We can use public text files to get going, even though it unfortunately means that votes are all public. For later rounds, PGP looks good. The PGP spec is public, and we happen to have all the required encryption algorithms.
Hmm, by the way. 4 out of 5 of the listed authors of our cryptography module are not included on Squeak People. Thus, it isn't only *my* collaborators who would be left out by "just use Squeak People".
-Lex
Hi Lex :-)
On who votes:
Like I said the last time you posted this list, the barrier against these people under SqP is practically non-existent. Just make them users, certify them, and send them their password. With about 20 minutes of your time, the problem is solved. Let me know, I'll certify people I know from that list.
Not everyone you know is on SqP, but its very easy to fix - probably easier than if Alan Kay forgets to apply on a wiki page. So "people x,y,z are not on SqP" doesn't look to me like a valid reason to reject SqP.
I don't think we need to get everyone rating like mad, seems the algorithm is pretty sensitive, and we can tweak that later on. Anyway, I predict over 200 new ratings within 24 hours of the moment we state that this is the criterion for voting.
So my gut feeling is that SqP will take far less effort to implement than what you propose. So, I know we're not alone on this list... anyone else have an opinion either way or some other way?
On who proposes referenda, a seconding mechanism would be a fine alternative. So one person proposes, one seconds? Proposer gets to set the date?
And a critical matter - do we need a lower bound on voter turnabout for a proposal to pass? if not, we probably want a lower bound on voting time.
Daniel
Lex Spoon wrote:
It's barely possible that the Squeak People designations look strange to me simply because all of the people I am thinking of really have left Squeak. But let me take one more stab at showing you how strange it looks to me. Here are the people I posted before, this time with annotations. None of them will be allowed to vote if we "just use SqueakPeople". Don't you guys agree that leaving these people out is rather bizare?
Mark Guzdial, who wrote PWS and PWS/Swiki, who taught a Squeak-based university class to thousands of people over the years, who sponsored several Squeak-based Ph.D. projects, who authored a text book based on Squeak, and who co-edited the nu blue book.
Bolot Karimbaev, author of Comanche.
Alan Kay.
KK Lamberty, a Ph.D. student who did the DigiQuilt project in Squeak.
John Maloney, who among many other things brought Morphic to Squeak, gave Squeak a sound infrastructure, and is now working on the Scratch project.
Jeff Pearce, who brought "Alice" to Squeak.
Jeff Rick, who put many hours into building and refining ComSwiki, and who has put in years of scholarly work on wikis.
Kim Rose, who has been with Squeak Central since the beginning, is a major player at ViewPoints, who co-edited the nublue book with Mark Guzdial.
Jim Rowan, a Ph.D. student who did multiple projects in Squeak.
Nathanael Shaerli, who gave Squeak the Genie gesture-recognition system and who worked out the traits system of restricted multiple inheritance.
Instead of rejecting all of these guys, here is a proposal based on the previously posted list of Squeak-based criteria.
For the first round of membership, let people apply on a Wiki page and post their reason for being considered a member. Objections and requests for more information can be posted in Wiki style. After a month or so, I expect that a consensus will emerge on practically all applicants (and, in fact, likely 100% of them will be included), and that will be the initial membership.
This seems easier, on the whole, than trying to invigorate the entire community to actually take part on Squeak People.
For later rounds of membership, I propose the same kind of thing, except that instead of having public review, we have one person in the group review them carefully and then post their findings publically somewhere. If no one objects after a certain period of time (one month?), then a person can then be included in the Squeak club with full privilages.
Again, this seems easier on the whole than trying to get people to maintain their Squeak People ratings.
For referenda, limiting who can propose them means that power gets rather centralized. Allowing resolutions to be proposed by anyone is a helpful check against the leadership getting carried away against the wishes of the group. Instead of defining an inner circle and using that to limit referenda proposals, how about requiring proposed referenda to be seconded? There are easy ways to avoid trivial referenda if that becomes a problem, but once power is centralized it is really hard to move in the opposite direction.
Daniel's non-functional requirements look excellent to me. We can use public text files to get going, even though it unfortunately means that votes are all public. For later rounds, PGP looks good. The PGP spec is public, and we happen to have all the required encryption algorithms.
Hmm, by the way. 4 out of 5 of the listed authors of our cryptography module are not included on Squeak People. Thus, it isn't only *my* collaborators who would be left out by "just use Squeak People".
-Lex _______________________________________________ Elections mailing list Elections@lists.squeakfoundation.org http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/elections
elections@lists.squeakfoundation.org