So practically you’d need to have some sort of a grace period for children of a deleted node. Like, “Your parent node has been found by our board to be actual human sewage. Your account is suspended. You have 30 days to find a new parent or permanent account deletion will occur.”
I was going to say that a fair number of Nazis are crypto enough to fool okay people into being their friends, but then I realized one, Nazis usually have other shitty things about them that ought to be red flags to okay people, and two you addressed it with the adoption mechanic.
Ok, sure. Someone who is just a non-woke gamer gets an invite. They are careless with their invites, and invite a nazi. The nazi posts a nazi thing. Then the whole branch is cut off the tree. The point isn’t to avoid anyone ever posting anything bad, that’s impossible. The point is to err on the side of keeping bad actors off the network, even if good actors are an occasional casualty. Much much better than current networks which err on the side of maximizing the number of user accounts.
The Tree is interesting. Do we think it would work on scales bigger than manual moderation?
Only one way to find out.
PROJECT ST approaching
Also, part of the hypothesis is that the graph greatly reduces the burden of the manual moderation. Imagine the case of the bot army.
On Twitter you have to detect and ban each and every individual bot account.
On the hypothetical graph, it would be very easy for one user to create an army of bots, but that bot army would all be part of that single user’s branch. It will be trivial for one human moderator to eliminate the entire army of bots in a few clicks.
In theory it should be extremely difficult to get bots that are horizontally spread out along the tree on may separate branches. You would have to convince many different people in many different communities to give you invites so that your bots were not recursively related. I’m sure somone will find a way, but I can’t think of how to do this easily or at scale.
You could scale up or down how Tree your app is too. It could e.g. just be a membership model for a forum, or you could make Next Twitter with the sibling/parent/child interactions from earlier.
Yeah, I think it might be easiest to just use the tree for membership. Once people are in, it can just be Twitter. The only difference being that you auto-follow the person who invited you and anyone you invite. We can also let people browse the tree as well as suggest they follow people 2-3-X steps away on the tree.
The main feature of your tree model is easy purges. That’s not a feature most people care about.
That’s a good question. It seems to be the number one feature that I see people requesting. They want to ban all the nazis and all the bots. But how many people is that really…
I think you are right, though, that there needs to be another feature that users actually want that puts it above the competition. Will the visualization of the tree, and discovering your social connections be killer enough? Facebook never lets users actually visually browse the social graph, even though they have it.
This reminds me of a system Couchsurfing used to have before they changed their revenue model.
There were three levels of account:
- Free account: free to sign up
- Verified account: Pay $5 with a credit card to confirm identity and address.
- Vouched account: The original creators of the site could vouch for people. Those people were given a limited number of vouches (between 2-5 iirc) with which they could vouch for a verified user. You didn’t get an official status unless you were vouched for by two independent accounts. Additionally, there were social rules about the vouching, like limiting it to people you had met in person, people who already maintained a certain level of participation in the community.
It was interesting… being vouched for was very rare – Phil received one vote but not a second before the changes happened. I think some of the variables needed tweaking to make people a little less precious with their vouches.
Also, my TOTD
Does this really need to be a social network in-and-of-itself? Couldn’t you just have this as the subscribable whitelist Rym’s always talking about? Just have it sit on top of all the regular social networks (I know, far easier said than done)
That’s… not a bad idea at all. The problem is how to do it given the existing APIs.
Take Twitter for example. Twitter doesn’t work with a white list. It works with a bla(o)ck list. You would have to first somehow block every Twitter account, and every new Twitter account. Then only unblock the ones in the tree.
I guess you could make your own custom Twitter client that does the blocking client side, but that’s not great. Twitter is also not great for third party clients these days.