Edit Spam?
There is a paragraph at the bottom of the Context Box in this discussion that appears to be spam. It has these keywords:
> visualizacoes youtube
comprar visualizacoes youtube
como conseguir visualizacoes
Seguidores instagram
youtube views
como divulgar video youtube
como ganhar dinheiro com youtube
Followed by a link.
Thanks for bringing this up @strypey ! I think we need to distinguish the editing of a discussion's title and context from the editing of comments inside it. I'm pretty amenable with a system in which thread comments get 'locked' after a short while, although IMO an ideal system would allow users to attach "Edit Notes" to their comments based on subsequent discussion. (For instance, something like "EDIT: Thanks to Jane Doe, I now realize that this comment was based on a false assumption.") I currently use a similar "edit policy" for my comments in discussion groups-- wherein I leave my original comment intact.
Regarding discussion titles and context: Loomio has helped me to see the value in leaving these indefinitely open to edits. However, I think that's only feasible when the edit history is easily available to everyone. And, I don't know why I can't currently see the edit history for the Loomio thread context boxes (using the angular interface.)
There are other threads about comment editing, so I'll leave that to those. Sometimes the context box edit history doesn't show up because the context box in question hasn't been edited yet ;) If you are sure it's been edited, and you can't see the link to the edit history, that's probably worth filing a bug about.
Coming back to the topic of this thread, anything need to be done about the edit spam I mentioned? Is there a vulnerability that needs to be patched, requiring a bug report? Is it just a case of spam user management on Loomio.org? Something else?
I had to switch back to the old Loomio interface to see the context box history. Screenshot attached. I'll report the fact that I can't see context box history in angular now.
I'm guessing that the IP of the deactivated account could be identified and banned by Loomio admins. Of course, that wouldn't eliminate all possibilities of further exploitative behavior by the same person.
I'll go into some detail on this issue below, because 'drive by exploits' are a personal focus for me as a discussion group engineer. However, I wouldn't suggest that this exploit issue should be a pressing developmental priority. The issue would be much more urgent for any super-popular Loomio groups we may see in the future.
Ultimately, group coordinators should have tools which enable them to add unknown people to groups with a minimized risk of 'drive by exploits'. Reason: If we only invite known people to a discussion, as with a purely 'trust-based' group, our potential for collective intelligence suffers. Sorry that there's no TL;DR summary for the analysis below; I hope it's worth reading. (Also sorry for the unnecessarily large image attachment above; I was rushing and it's too late for me to edit it.)
Ideally, the personal 'cost' of creating a new account and getting the right to edit a context box should be high enough to effectively minimize clickbait spam. We definitely don't want account creation to be tedious, so I lean towards options for 'permissions control' and a 'progression system' for account holders.
Generic permissions 'levels' could be as simple as having an option for two 'classes' of group participants: guests and members. Guests would have opportunities to provide valuable content, but their posting frequency may be limited, and they may not be able to do certain things-- such as edit context boxes.
Permissions control management could be as simple as admitting new participants as 'Guest', by default, but allowing coordinators to promote them to "Member" at any time. (Perhaps immediately, for well-known parties.) Such a simple, manually controlled system would put the responsibility on the coordinators (and hopefully, an empowered community too) to create good practices for the promotion of unknown guests to member status.
Alternately, or supplementally, group coordinators could get an option to activate an 'automated progression system'. There are various ways to do this. The basic idea is: if someone participates substantially in a group for a substantial period of time, they are automatically promoted to a new level of permissions. Such an automated system could, of course, reduce potential 'work load' for coordinators of large groups.
The key to any automated system would be to minimize the opportunities for minor exploits by new guest accounts. This is definitely possible, but it requires finesse in any system which allows super-easy account creation. (And I certainly favor such systems.)
I can imagine Loomio developing a many-to-many system of roles and abilities for group users (who can post proposals? Who can comment? Who can edit their own / other people's comments? Maybe [although this is potentially hairier] who can vote?), instead of the handbrake on/off by group settings we have now.
This would be, by design, super flexible, with sensible defaults.. so you'd be able to define what a 'Member' can do, or call them 'Padowans', or whatever.
It's on our mind, but it's a little ways down the list in my head, since I'd rather spend a whole bunch of time making it really easy to make decisions with a group, and then focus on experiential things like permissions and coordinator experience.
Apologies for no TL;DR from me either :}
I'm really interested in this balance between ease-of-access and spam-prevention, not only as it affects Loomio, but as it affects every online collaborative service. It also complicates every kind of federation of social software, and especially federated identity.
Comments can be deleted by coordinators, which is probably why I've never seen spam comments, although I would like for deleted comments to be noted, and linked to the original comment, visible only to members (for transparency and accountability of coordinators). Perhaps it should be a coordinator role to update Context Boxes? Or the role of the person who started the discussion, although that wouldn't prevent spam accounts with member permissions from starting discussions.
Generic permissions 'levels' could be as simple as having an option for two 'classes' of group participants: guests and members.
Another way to express this in organisational language is "associate members" and "full members". Your proposal is elegant in many ways, but I am nervous about going beyond three levels of permissions (admin, coordinator, member), and not only because it gets complicated to code, and for users to understand. There is danger of erecting bureacratic pyramid heirarchies, and losing Loomio's founding value of horizontal decision-making.
Preventing bots from creating accounts would also help, although it's a challenge to create captchas that don't discriminate against blind users.
The user who started the discussion appears totally legit, so I'm suspecting this spam has been added in a later edit. IMHO vulnerability to spam is another good reason not to allow comments to be edited indefinitely.