[WikiEN-l] Accountability: bringing back a proposal I made

[Gwern Branwen]

Date: Mon, 05 Mar 2007 16:07:11 -0500
In-Reply-To: <45EBDB9F.2070808 at wikia.com> (Jimmy Wales's message of "Mon, 05
	Mar 2007 17:58:07 +0900")
Message-ID: <86r6s3wgao.fsf at elan.rh.rit.edu>
User-Agent: Gnus/5.110006 (No Gnus v0.6) Emacs/22.0.95 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain; format=flowed
--text follows this line--
Jimmy Wales <jwales at wikia.com> writes:

> In response to the EssJay scandal, I want to bring back an old 
proposal 
> of mine from 2 years ago for greater accountability around 
credentials:
>
> 
  http://lists.wikimedia.org/pipermail/wikipedia-l/2005-May/022085.html
>
> At the time, this seemed like a plausibly decent idea to me, and 
  the 
> reaction at the time was mostly positive, with some reasonable 
  caveats 
> and improvements:
>
> 
http://lists.wikimedia.org/pipermail/wikipedia-l/2005-May/thread.htmlto 
read the entire thread of "An idea".
>
> Nowadays, I bring back the proposal for further consideration in 
  light 
> of the EssJay scandal.  I think it imperative that we make some 
positive 
> moves here... we have a real opportunity here to move the 
  quality of 
> Wikipedia forward by doing something that many have vaguely 
  thought to 
> be a reasonably good idea if worked out carefully.
>
> For anyone who is reading but not online, I will sum it up.  I 
  made a 
> proposal that we have a system whereby people who are willing to 
verify 
> their real name and credentials are allowed a special 
  notification. 
> "Verified Credentials".  This could be a rather open ended 
  system, and 
> optional.
>
> The point is to make sure that people are being honest with us 
and with 
> the general public.  If you don't care to tell us that you are a 
PhD (or 
> that you are not), then that's fine: your editing stands or 
falls on its 
> own merit.  But if you do care to represent yourself as 
  something, you 
> have to be able to prove it.
>
> This policy will be coupled with a policy of gentle (or firm) 
> discouragement for people to make claims like those that EssJay 
  made, 
> unless they are willing to back them up.
>
> How to confirm? What counts as confirmation?  What sorts of 
things need 
> confirmation? These are very interesting questions, as there are 
  many 
> types of situations.  But one thing that we have always been 
  very very 
> good at is taking the time to develop a nuanced policy.
>
> Just to take a simple example: how to verify a professor?  This 
strikes 
> me as being quite simple in most cases.  The professor gives a 
  link to 
> his or her faculty page at the college or university, including 
  the 
> email there, and someone emails that address to say "are you 
  really 
> EssJay?"  If the answer is yes, then that's a reasonable 
  confirmation.
>
> We can imagine some wild ways that someone might crack that 
  process 
> (stealing a professor's email account, etc.) but I think we need 
  not 
> design around the worst case scenario, but rather design around 
  the 
> reasonable case of a reasonable person who is happy to confirm 
> credentials to us.
>
> (This is a lower level of confirmation than we might expect an 
employer 
> to take, of course.)
>
> For someone like me, well, I have an M.A. in finance.  I could 
  fax a 
> copy of the degree to the office.  Again, someone could fake 
  their 
> credentials, but I don't think we need to design against some 
mad worst 
> case scenario but just to have a basic level of confirmation.
>
> --Jimbo

I like this proposal, as it is similar to one I've been bruiting 
about for the last few days.

The way I see it, there is no really golden mean between full 
pseudonymity (where you give few to no details about who you are; 
where "few to no" means that the obtainable information is limited 
to basically the sort of stuff userboxes cover - excluding stuff 
like your real name, address, phone number, employment or 
employment history and other things like credentials), and full 
transparency. When I saw full transparency, here I mean that 
enough information about one's life is  given that, in principle, 
one can verify claims about expertise, official credentials, and 
so on.

Now, note that I emphasize verifiability. This differs from 
Jimbo's proposal. If I may use some out-moded descriptions, 
Jimbo's idea where one has another rank in the hierarchy where 
credentials are deemed verified (past tense) is more of an 
Immediatist/Deletionist sort of proposal. I am not surprised at it 
- it is a truth of Wikipedia history that every time a scandal of 
some sort scars the community, restrictive I/D proposals pop up 
and gain credence (example: disabling anonymous page creation 
after Seigenthaler), but that doesn't mean we should just do 
them. If we implement any such proposals, disarrayed and dismayed 
by a recent scandal, we will have to live with it a long time, 
particularly if orders for it come from the top and are enforced 
by changes in the MediaWiki software (remember, "code is law"). I 
understand many elder editors are not particularly convinced 
restricting page creation worked and that it should be turned on 
again, but because the order for that restriction came from the 
top and is implemented in code, it is literally impossible for 
anons to create pages except through meatpuppets of registered 
editors. And this has been the case for quite literally years 
(come this December, I think it will have been 2 full years). So 
if we are considering proposals which will give a subset of 
registered editors official imprimatur, possibly reflected in the 
software itself (not sure how else the "Verified Credentials" 
would be implemented), then we need to be rather careful.

That said, I also don't like the proposal because it adds yet 
another level to the hierarchy (how will it go? Anons, editors, 
Verified Credential editors, admins, bureaucrats, Checkuser, 
Oversight, stewards, board, Jimbo?), because it centralizes the 
activity of checking credentials, and just in general imposes 
overhead - the whole point of wikis is to reduce overhead and bet 
that it enables good users more than bad ones. 

What we need to do is discourage the in between. There is no 
golden mean, but this does not mean we should set up yet another 
heavyweight process to verify people on the transparent end of the 
spectrum, but rather we culturally or perhaps by guideline or 
policy say that claims to expertise not backed up by verifiable 
information should be discounted and the claimer treated exactly 
as if they were pseudonymous.  Let people claim to be professors, 
if they want, but let no one treat them as professors without 
sufficient reason to believe that. One cannot legislate common 
sense, as the saying goes.

-- 
Gwern
Inquiring minds want to know.