Douglas_Knight comments on Open thread, September 2-8, 2013 - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (376)
Is there a good way to avoid HPMOR spoilers on prediction book?
I used feed43 to create an rss feed out of recent predictions. Then I used feedrinse to filter out references to hpmor resulting in a safe feed. (Update: chaining unreliable services makes something even less reliable.)
You could do the same for the pages of recently judged or future or users you follow. I think feedrinse offers to merge feeds (into a "channel") before or after doing the filtering. But if you find someone new and just want to click on the username, you'll leave the safe zone. Even if you see someone you have processed, the username will take you to the unsafe page.
A better solution would be to write a greasemonkey script that modified each predictionbook page as you look at it.
The final feedrinse feed works in a couple of my browsers, but not chrome. Probably sending it through feedburner would fix it.
feed43 was finicky. The item search pattern was:
<li class="prediction{%}">{_}<p>{_}<span class='title'><a href="{%}">{%}</a></span>{%}</li>
The regexp I used in feedrinse was /hp.?mor/
It is case insensitive and manages to eliminate "HP MoR:", "[HPMOR]", etc. It won't work if they spell it out, or just predict "Harry is orange" without indicating which story they're predicting about. In that case, someone will probably leave a hpmor comment, but this doesn't see such comments.