Re: [UG-ADMINS] Re: ZendCon Panel and Next Steps
[email protected] ("Rafael Dohms")
| Newsgroups | ug.admins |
|---|---|
| Message-ID | <[email protected]> |
I'll respond below inside you mail Rafael Machado Dohms http://www.rafaeldohms.com.br On Sat, Oct 25, 2008 at 23:43, Tim Stiles <[email protected]> wrote: > I've done a couple of sites that needed the cooperation of multiple > non-professional groups. You must plan for the fact that the groups are > chaotic. The groups persist, but the contacts for those groups change > frequently, without notice, and almost always without any sort of plan for > smooth transition. They also are rarely interested in your problems. > Sometimes, the group entries are just abandoned, but there are sometimes > disturbingly loud conflicts and power struggles over them, which have > absolutely nothing to do with you or your purpose. This can throw your > simple contact list mechanism into complete chaos. > A controlled state is nirvana in any situation, so yes.. chaos is the default result. But we can have controlled chaos. > > A) I know that fully automated systems are what we all desperately want > (since our real lives always get in the way of the fun stuff), but it almost > always results in entries that are obsolete or deliberately deceptive. > Requiring someone's conscious approval before an entry comes online is a > strong deterrent to misuse and almost always guarantees validity coming in. > Once they are in the system, there is less chance that people will play > games, automate everything you can. If you have to automate the process of > approving entries, make certain there is a responsive dispute mechanism so > that dead or contested links can be pulled offline quickly. (Case in point, > I've been trying to get an entry corrected on PHP.net's calendar for three > years, but I didn't submit it, he's off the grid now, and an update protocol > is practically non-existent.) This is an interesting point, and i will implement a dispute system so we can solve these problems, but without getting into them, which is negative and we just would not want the hassle. And as we here in Brazil have 2 or 3 coordinators in each group, not just one, i'll implement something along that line, so we don't rely on a single contact. > > B) Active groups are proactive about promoting their existence. They won't > mind responding to a periodic email asking "Are you still there?" It's also > easier to get them to update their info if they don't have to remember to > visit and log in to your site on their own. An authoritative email contact > is mandatory, a secondary email contact is preferred. Instead of an > entirely login driven system, make use of a timed-access email-invitation > system. > This is not rulled out. We have a automated verification system, which i'll explain in the next step. But on top of that we have an administration area for the groups where i thought of putting a "I'm alive" button for pro active verification. The email will also happen in case the AVS detects lack of activity, in which point 3 sequenced alert mails will be fired for the admins to reply with an "hello" Also the authentication module will be .. well modular. So we can plug new ways to log in as we see fit, the first prototype will go with a WP integration as its our current reality in the national representation here in Brasil. > C) PLEASE PLEASE Make entries expire unless they are confirmed via the > people who submitted them. Don't rely on a web page test mechanism - web > pages can stagnate for years, transfers of authority and administrative > access to those sites are not always seamless, and internal disputes within > groups are not uncommon, let alone unheard of. Set up plans for changing of > administrative contacts, especially the circumstance where the original > registrant drops off the face of the earth with no warning. (example: if a > group leaves meetup.com, any other meetup.com registrant can claim their > name with no warning. I've seen it happen with a couple of different web > design groups in the Dallas area.) > I will implement a verification system. The automated Activity Verification system (AVS) is a X step program. I'm creating various mechanisms for activity verification, this means, checking the website, checking the mailing list, php calendar maybe, and this will be extendable. ehehehe I just spent a few hour explaining this in our list, so let me try to summize. The process is composed of a list of "criteria" ordered by relevance. So for example, site activity, list activity..etc... The robot will periodically run through this list checking for the correct metrics for each criteria If any criteria is met.. it exits, if not it goes on to the next criteria.. so if a groups has X posts on the site in a month, it passed and we go on to the next group, if it does not we go check the mailing list.. and so on. If all criteria dont match, we sent a warning mail.. 3 chances for the admin/responsible to hit the "i'm alive" button, if he does not, then the group goes inactive waiting for re-activation. (never delete, just soft delete) So this balances manual and automatic verification and is totally customizable from admin panel.. like how many mails /month is active? how many posts? and such... Well this is what i'm planning and has been approved in brasil, so once its in a beta release... i'll put it up so that we can look at it from a international point of view and see if it fits... and if needed we will branch it.. do a php.net branch and adequate it to other standards. But i think we have most of the problems covered. And i'll incorporate any new suggestions as they come. Thanks for the feedback guys.. good stuff! > > BAD DATA IS FAR WORSE THAN MINIMAL DATA. > > Tim Stiles, > Co-Organizer, DallasPHP > WatchMaker, Icomex.com > > > > On Oct 25, 2008, at 5:16 AM, Rafael Dohms wrote: > > I have incorporated some of the suggestions given here, but i've added a >> automatic layer on top of it, so that most of the time no human >> interaction >> will be needed, but we will still have the humam factor for lat case >> scenarios, before we pull the plug. >> > >