[WikiEN-l] verifiability

[Conrad Dunkerson]

steven l. rubenstein wrote:

> I agree.  Do you have any concrete suggestions?  You and a number of other 
> people have made some fine points; perhaps we can move this discussion to 
> the Verifiability talk page and start discussing how to put these ideas 
> into practice?

The only method of having 'verifiability' which I can think of that 
might work without extensive changes would be to have a way of marking a 
particular revision of an article as a 'verified' copy. Since verifying 
facts is already part of the 'Featured Article' process it would make 
sense to mark articles as 'verified' when they become featured and other 
articles could be 'verified' as needed. Then someone could easily 
compare the current revision to the most recent 'verified' version and 
see if there are any radical changes of fact.

Obviously this 'verified flag' couldn't be open to the general public or 
the vandals would just mark their vandalism as 'verified'. One way to do 
this without making any system changes might be to have a 'Verification' 
account (accessible only to highly trusted editors) which is used only 
to put 'verified as of <date>' tags on articles. That accounts' 
contributions could then be viewed to quickly identify articles which 
have been changed since verification... article histories could be used 
to show diffs between the current version and last verification edit, et 
cetera.

Of course, more robust tools for this sort of 'verification' tagging and 
tracking could be developed with coding changes to the system, but with 
a bit of thought/effort it could be implemented under existing code. 
Putting an 'unverified' tag on every article and 'priority for 
verification' on important ones would allow the creation of easy lists 
for people to go through and verify... likewise lists of 'most edits 
since last verification', 'requests for verification', and suchlike 
could be developed.

The essential element here is a way of identifying a version of each 
article as 'accurate' for later reference. Without that you'll always 
have new edits coming in and no way for those unfamiliar with the 
subject to tell if the old or the new is more likely to be accurate 
without research.