"Zope-certified Python Engineers" [was: Java and Python]

[John J. Lee]

On Thu, 21 Mar 2002, Laura Creighton wrote:

> > ( Or maybe this begs for a Slashdot like system to
> > distribute the review, grading and selection of projects. )
[...]
> You want something more like the scientific reviewing of a technical
> paper than slashdot's 'all you need to vote is a pulse'.  Otherwise

This will not happen, as I'm sure you realise.  :)  Who would do it??

> in your selection committee).  The best way to do this is to provide
> a certain amount of meta-data with your assessment -- so we can
> find out not only that Steven Majewski thinks that XXX is cool, but
> also that Steven Majewski has a reputation around here, and most
> people think that Steven Majewski's opinion of something is valuable.
[...]
> I think that this sort of metadata belongs in the sort of
> cataloging system we should have to implement PEP 262, as well.
> Of course, I am biased -- AB Strakt is building something to do
> this, because we think that its a huge world problem that needs
> fixing.

There seem to be two ways to search for documents in a useful way, both of
which are useful.  One is to have lots of carefully and expensively
maintained metadata.  The other is to self-consistently rank the documents
by looking at the links between them, as does Google (one of the wonders
of the modern world, IMO).

I think Google has demonstrated that ranking is better done using the
self-consistent approach (though perhaps a little bit of metadata added to
this approach would work even better? -- for example, Google has no
conception of author AFAIK).

In general, and for a Python archive in particular, I guess there are two
problems: which metadata, and which software.  So, what's wrong with the
web + Google (not saying there isn't anything wrong, just interested in
how you would improve on it)?


[OT: A question I've asked many times and got no answer to is 'why has
nobody done self-consistent ranking for academic papers'?  I fear the
answer is simply that most of the data is owned by a small number of
companies who, as a result, have no real incentive to advance the state of
the art.  Still, maybe somebody out there is trying to do it...]


John