[TYPES] Two-tier reviewing process

[Matthias Felleisen]

On Jan 29, 2010, at 1:41 PM, Alan Schmitt wrote:

> On Fri, Jan 29, 2010 at 6:21 PM, Kim Bruce <kim at cs.pomona.edu> wrote:
>> [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list 
>>  ]
>>
>> I agree with Matthias that we place too much emphasis on conference  
>> submissions, and agree with those who recently have been making  
>> more of a plea for computer scientists to move more toward journal  
>> publications.  I usually tell people that conference submissions  
>> are refereed for interest (novelty, impact, etc.) and "likely  
>> correctness" -- with high variance in acceptance due to a  
>> restricted pool of referees and the time pressure.  On the other  
>> hand, journal publications are refereed with more weight on  
>> correctness (after all, these are usually the only places you get  
>> to see the details necessary to verify correctness).  We should be  
>> moving to a hybrid system where results are announced at  
>> conferences, but are only considered settled when they appear in  
>> journals (with several conference papers likely being coalesced  
>> into a more significant journal article).  I'm not saying we should  
>> move away from refereed conferences, but they should not be the be- 
>> all and end-all.
>
> Unfortunately journals often require new results. I have tried several
> times to published coalesced papers with proofs, and invariably I was
> asked where was the new, unpublished result. But maybe I was unlucky.

It's much more likely that the culture of publication is all
wrong already. People may lack the time, training, tools to
perform the journal reviewing tasks properly -- and they may
just lack the incentives given our emphasis on the haphazard
conference process.

To some extent, it's a chicken-and-egg problem, which is why
it is so difficult to change things. -- Matthias