To answer your first question:
I'm using a custom scraper to get the results. I request the job pages through a random http proxy every so often, parse them, and update the cached json of all the comments. Getting a reliable, free, machine-friendly list of http proxies is harder than I thought. It was automatic, but the proxy provider I was using went down (doh!) so now I manually pull one from a list and trigger the scrape.
On the server I'm using a tiny express app (node.js) + redis to store hidden posts per use based on a cookie.
On the browser just a little bit of twitter boostrap + jquery + some custom UI helpers to generate the page. The searching and hide/show is done within the browser itself. I try to involve the server as little as possible.
To answer your second question:
It will update as soon as I kick off another scrape. I hope to re-enable the automation on this soon so it will only lag 10-15 minutes behind the actual page. Right now it could be quite a while (up to 12 hours).
Yeah! ctrl-f was tough. Especially as the comments spanned multiple pages and weren't sortable by date.