The approximate score will with accurate edit distance give two equal
contributors in the first case, it will credit A with all edits if a
limit is set on tracking of minor edits that is anything above zero
with no edits on B, and it will credit the last of two editors if last
contributions wins. There are no really good approximations, and t
doesn't help to use "percentages".
Best I know of is dimension reduction and measuring path length, but
that too fail for some combos of vandalism/reverts.
On Fri, Aug 16, 2013 at 5:08 PM, David Cuenca <dacuetu(a)gmail.com> wrote:
On Fri, Aug 16, 2013 at 10:58 AM, John Erling Blad
<jeblad(a)gmail.com> wrote:
We don't use votes... ;)
I think Lydia was referring to the votes on the bugzilla bug page, 14 votes
so far :)
If we forget about the implementation of badges and discuss the
contributions; there are no single correct way to weight
contributions. Assume some user A write N characters as a continuous
string, and some user B writes the same number of characters spread
out over a text changing N words into something else. Those two edits
can have the same edit distance but still have a completely different
entropy. In the last case, who "owns" the changed words? The original
author or the later one? This isn't obvious at all.
I thought Wikitrust [1] and others [2] had already addressed this issues?
In any case there is no need for an "exact" attribution, an approximate
percentage score would be a good enough solution for practical purposes.
[1]
http://www.wikitrust.net/
[2]
https://www.mediawiki.org/wiki/User:Mshavlovsky/Authorship_Tracking
_______________________________________________
Wikitech-l mailing list
Wikitech-l(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l