[GET-dev] web hits ranking

Tom Clegg tom at tomclegg.net
Thu Jun 24 13:22:11 EDT 2010

> Are you familiar with the webhosting "link exchange" strategy?

IMO our decision to advertise/link to x, y, z should be completely
independent of whether they [intend to] link to us.  Much better to
select them based on how useful they are to our users.
Philosophically, but also because I assume search engines do their
best to distinguish "link exchange" links from "honest reference"
links.  If we need some more publicity, I'd be more inclined to use

In any case, based on Trait-o-matic's search index performance, I
predict GET-Evidence will do extremely well on Google et al. based on
content alone.

I do like the ideas about providing incentives for "all-star
contributors" and "all-star referrals".

More ideas for collecting info to rate editors (with basically random # values):

* curator accepts edit: editor gets +1 (basically "non-spam edit")
* user endorses edit: editor gets +10 (e.g. curator hits "non-trivial
edit" when accepting?)
* user endorses page: each editor gets +5
* user endorses editor: editor gets +50
* max +score for a single article: 50

AFAIK the reason robots.txt is still there is that we don't yet have
"curator" functionality (i.e. newcomers shouldn't be able to post
publicly-viewable spam).  So, we should focus on doing that in such a
way that editors accumulate points as the edits are accepted.  Then
add other ways of accumulating points.

I think it would also be helpful if editors (who have had edits
accepted) can provide some basic info like affiliation or an "about
me" page, so users can get a better idea of whose stuff they're

(Ideally all openids would *be* "about me" links, like Madeleine's...
but afaik there is no way to achieve that with Google, for example.)


More information about the Arvados mailing list