[e2e] was double blind, now reproduceable results

[Jon Crowcroft]

ah, but you miss the beauty of my scheme - we dont need to check EVERY
paper submitted 

we can do Random Early Debunking

as the submission rate hits a threshold, we start to pick papers at
random and apply the test (maybe this is Fair RED), and if our pool of
handy grad students can't get the same result, we start dropping
papers from this source....

just the existence of this should is sufficient to make most people behave


of course some one will launch a DDoS attack on our checking system -
but all is not lost - we have many grad students, and they can do more
classes checking the submitted experiments work, leaving those of us
with succesful grad schools to submit more carefulyl checked papers

now, this does leave the question of who will check that the grad
students have re-run the experiment , and not just thrown a dice and
checked the pass/fail box accordingly - this is of course easily
solved, but I dont have space in this email to put my elegant proof

In missive <20040520051018.A40786 at xorpc.icir.org>, Luigi Rizzo typed:

 >>it would be nice if it could work, but probably not in this world...
 >>if the policy is enforced it may improve the quality of work, but
 >>to enforce it requires reviewers to actually check that the tools
 >>and data are sound -- if people have no problem in submitting
 >>half-baked papers, why would they behave differently with all the
 >>rest ?  So the work for PC members would actually increase a lot.
 >>
 >>> and the number of papers needed to get tenure wil, on average, be
 >>> reduced too since the impact of each work will increase.
 >>> 
 >>> who loses by this?
 >>
 >>conference organizers, travel agencies, pubs, companions of
 >>delegates, ... :)
 >>
 >>	cheers
 >>	luigi

 cheers

   jon