Go on over to the New York Times home page and view source. Drop your jaw at the HTML comments that reveal their gnarly SQL they use to generate their home page.
Each three article section has the SQL used to generate it in a comment above it. The funny thing is that the SQL is longer than the results.
select oid from assetrelationtree where otype='NYT_Article' and nparentid in (select nid from assetrelationtree where nparentid=0 and oid in (select id from collection where id in (select distinct oid from assetrelationtree where nparentid=(select nid from assetrelationtree where oid=1004547 and nparentid=0) and otype='Collection') and subtype='NYT_Article')) and ncode <= '20060407' and oid not in (1125004343548,1125004284421,1125004335857, 1125004327602,1125004279763,1125004286523, 1125004276389,1125004338045,1125004296004) order by ncode DESC,nrank ASC