I scrape only the front page once every ~4hrs (minimum)

Nearlyfreespeech.net doesn't have cron yet, so I'm using onlinecronjobs.com. If you want to scrape it on demand just go to http://hnweekly.watdahel.com/index/scrape

But, if you only scrape the frontpage, how can you know the points for all the past stories which are not on the frontpage anymore?

It's not 100% accurate, but it's close. The reason I did the website is so I can catch up with the good stories that were submitted earlier during the week (I usually read during weekends).