FarmersWrap

No problem! But I realized the decoder highly depends on JSONTokener, you wouldn't want to double-check the json string nor change the logic of logging JSONTokener error message. Then I...

I think the "normalize" here looks into the whole ranked list, and then rescores. It compares between doc scores in the list. Then try to make them stay within (score...

In that case, I would say yes. But I haven't checked our implementation. I don't know if we sort the list differently when we have same scores.

@lilyjge Ok, now I understand this better. I realized that we actually [apply avg to normalized runs](https://github.com/castorini/anserini/blob/master/src/main/java/io/anserini/fusion/RunsFuser.java#L108) The change should be under RunsFuser.java, right before merging. Is this the preprocessing...

yea, thats right. For the fusion methods, do I need to add weight as a function? And maybe keep avg, interpolation staying there and delete normalize? We then will have...

In terms of implementation, I just want to be clear that you want them all, right? https://github.com/castorini/anserini/blob/master/src/main/java/io/anserini/fusion/RunsFuser.java#L144 Why am I asking is because I think weighted can replace the other...

> Yes, for arbitrary weights, you can rescale so everything sums to one. That's a minor detail... Let's see a PR and have something concrete to discuss? Sounds good! Thank...