Why not solve the real problem and set up a search engine for News.YC? This has other advantages such as controlling what shows up in the results, different ways to sort them, etc.
It's easy to do with open-source software such as Solr (lucene.apache.org/solr) or Hounder (hounder.org, shameless plug because it is our product), which powers the wordpress.com site.
Edit: it looks like someone already did this: searchyc.com.
Either way, I think that PG is the king of guy who would rather work on his very special design (staying in Arc-land as long as possible) than adopting systems where you no longer have control.
Letting Paul work on the JavaScript makes sense, in this case. PG has no control over client-side code, so he prefers to not bother with it.
I didn't say that PG would refuse to use any system that was developed in Java. What I meant that he would never would put himself in a situation where the application would depend on Java for development, or to add more features.
To integrate with Google, you talk HTTP, not Java. To adopt Lucene, you'll have to talk Java. There is quite some difference, isn't it?
