[Search-l] Open innovation (was Which does ...)

[Bryan Bishop]

On Sunday 06 January 2008, jer wrote:
> Absolutely, I totally agree and have been thinking about what various
>   forms friendly (or serious) competitions along these lines might
> take.  Alas, a lot of groundwork has to be put down first to enable
> such fun experiments, and we're finally seeing that happen :)

Hm. I have been putting some more thought into this, and as far as I can 
tell, all that we really need to do is make the scoring method 
modularized on all of the servers so that we can drop in team modules 
whenever we want, and then we need some way to let grub zombies 
download the (this part is important) *reviewed* team code. 

Heck, in the future maybe we'll automatically download scoring routines 
from Sourceforge SVN, have multiple projects that put up packages that 
use a scoring API from Wikia search/grub. And for the daring, a way to 
choose the nightly builds. ;)

But back to more practical matters: I see the need for a review board of 
algorithmic search analysts to approve incoming team code, the 
modularized software to allow drop-in team components, and then some 
way to proliferate the code across the grub network, maybe by automated 
bit torrent for seeding purposes? But that's it. All of it is doable.

I do not know anybody at Google or Ask or something, but I am sure they 
have a few insightful search analysts that would have a fun time 
sitting on a review board for a few days a year, reviewing team code 
for submission into the competition. 

I'd be willing to write some code to make all of this happen, but it 
looks like the first priority is getting the code in a fancy package. 
As long as Jimbo or somebody could arrange the review board for people, 
ah, already more knowledgable than me in search algs. :)

Happy hacking,
- Bryan
________________________________________
Bryan Bishop
http://heybryan.org/