There has to be a automated way allow PJ to turn on user creation without
having the downsides it currently does. For example, freeshell.org requires people to send in $1.
I don't
think that would work here, because I'm guessing a large group of the
troublemakers have lots of $$'s behind them. But perhaps we can
brainstorm something that would work.
I'm sure under the current
system, I could get at least 2-3 patents out of this post. So I hope this
counts as prior art for years from now when someone does patent the ideas above.
Just for completeness, do the above on a computer, do the above on a TV, do the
above on a video game console, do the above on a phone, do the above on a
portable device, do the above on a computer network, do the above on the
Internet, do the above on Turing machine, do the above on a set-top box, time
shift the above, space shift the above, do the above in 3-d.
I'm
thinking the primary requirement is that it be easy for PJ to ban someone and
have it stick. I'm also guessing that a second requirement is for PJ to not
have to retain any identifying information so no one can demand she turn it
over.
I believe the second requirement is easy. This is standard
password hashing on some text that identifies the user. It's one way, so
forcing PJ to turn over the hash would not harm anyone.
The first
requirement is tricky. Here's an idea that won't work, but might give someone
else some ideas.
When someone signs up, they need to provide CC
information and a physical address. Then one of the e-card to snail mail
gateways is used with the CC information to send an authorization code that the
user types in to complete registration. Some normalized subset of the address
is used to create a password hash for banning purposes and to prevent duplicate
requests.
The main problems I see with this are requiring a CC and that
I'm sure it will only work in a handful of countries. I also believe it would
difficult to setup an address normalization routine that wouldn't be easy to
defeat while not having too many collisions. This might cover enough so that PJ
could manually handle the rest. I don't really have a handle on the
distribution of users.
Anyone have any better thoughts? e-mail
addresses won't work since it's way to easy to set up new addresses. Some
places get around that by not allowing gmail, yahoo, etc addresses and then
banning the whole domain when something goes wrong. I don't think that would
protect against astroturfing.
Coming from a completely different
direction, here's an idea that would only require a complete rewrite of the site
software. Allow anyone to register. However, all new users start out with a
trust level of 0. At trust level 0, the user can do exactly what an anonymous
user can do. As their trust level rises, the user slowly gains the
functionality of a registered user. A user gains or loses trust by posting
articles that currently trusted users chose to approve of. The amount of trust
gained, would be a function of the amount of trust the approving user has. A
user could only gain a limited amount of trust for a single other user. All the
approvals would be remembered, so banning a user could optionally impact
everyone they approved of.
The major downside of the second idea, apart
form requiring a major rewrite of the software, is that it would be difficult
for people with honest opinions that differ from the majority to gain much
trust. This could serve to limit discussion of differing view points.
[ Reply to This | Parent | # ]
| 
