[oops, appear to have replied to Anthony only - gmail is getting very irritating over this...] On 14/03/07, Anthony wikilegal@inbox.org wrote:

On 3/13/07, Andrew Gray shimgray@gmail.com wrote:

...

On 13/03/07, Anthony wikilegal@inbox.org wrote:

...

On 3/13/07, Andrew Gray shimgray@gmail.com wrote:

...

I really don't see anything wrong with me footnoting a) as "was born in Such-and-Such<ref>Personal correspondence with the Wikimedia Foundation, June 17th, reference ABC1234567</ref>.

I disagree. There are lots of problems with this. 1) You're not the Wikimedia Foundation; 2) It's not a published source which can be easily accessed; and 3) it's not a reliable source even if it's true, as the person no doubt does not remember his own birth.

1. I forgot the minor detail that I was corresponding with the chap via OTRS...

I'm not really familiar enough with OTRS to comment on that part, then, although OTRS should probably be mentioned in the reference.

Basically, he wrote to the Foundation asking why his correction kept getting taken out and replaced with the faulty cite. I explained.

What I'd have liked to do would be to have gone in and replaced the correction, footnoting it as "information sourced from Joe Smith by email, file # 1445364"; What this would do is make it about a minute's effort for anyone with access to the mail system to then check that I'd got it right, that the guy was who he claimed to be, so on and so forth.

The problem is that we can't make it publicly accessible; this is mainly for privacy reasons - there's no guarantee they didn't leave nonpublic comments in the same email, giving their address is essential to make it "verifiable" but really not a good idea given the Sort Of Odd People we sometimes attract.

...

2. Does it have to be easily accessed by a random passer-by if the

Foundation can verify it? That is really the point of this discussion... stating it won't work because of this is kind of circular.

I see how it was circular. I do think the encyclopedia's references should be more than "trust me, I work in OTRS", but I guess that's not obvious.

FWIW, we're relying on this now for a good chunk of our copyright clearance.

I don't think it's at all silly. The fact that we're talking about a usually unimportant fact such as a birthday makes it excessive, perhaps, but I think there's far too much acceptance on Wikipedia of facts which are altogether unsubstantiated. If all we know is that you claim that your mother claims that you were born on X, then IMO we shouldn't state as a fact that you were born on X.

I think we're getting into pretty tangential issues about the nature of verifiable fact, here...

...

Okay, one we might have an interest in getting right... manner of death. We usually give this if known. Not at all unknown for obituaries to get it wrong, especially if published quickly; small details in obituaries are rarely corrected afterwards for various reasons (most often that the only people who know it's wrong are otherwise preoccupied with mourning), and all too often that's the last thing published on them before we come along.

Spouse's names, that's another one we get a good few corrections over. Little things, yes, and we can say "why should we include them?", but the fact is we *do* include them, and it really seems futile to insist on a method whereby if we include them we get them wrong.

By all means we shouldn't be getting things wrong. If a reliable published source says that Bob died in Queens and Bob's mother calls up and says he died in Brooklyn, then we certainly shouldn't say in the article that he died in Queens. But it shouldn't say that he died in Brooklyn either.

The problem is negative sourcing. When we can't provide an opposing cite, then the Queens detail will keep creeping back in...

...

Which this proposal is one means to achieve. Not everything someone quibbles with the article over is significant, it's just that when it is significant we're less happy to take their unadorned word for it...

I think this is too much of a departure from the usual way of Wikipedia. But maybe it's something which could be done in a separate project, and included in Wikipedia by reference. A project for original research, maybe?

A very clumsy method would be WMF publishing a sort of weekly "errata sheet", for people to incorporate...