What does "open source" mean in this context? I think they just mean they will publish the algorithm somehow? Are they actually releasing code under an open license?
It may seem pedantic, but the root of my question is that when I read the article, I don't actually understand what it is they are going to do.
(Also, from what I understand, there is lots of manual adjustment going on. So if they publish an algorithm for how tweets are ordered, or weighted, that would be before subjective manual adjustment anyway, making it at partly meaningless, no?)
I'm not one to vouch that security through obscurity is good, but knowing how the algorithm works, depending on how its designed seems prone to abuse. If what they release is not specific to avoid the potential for abuse, I don't see how useful knowing it will be. If it ends up like I believe it will, all the useful stuff will be in the weights and models that run it, which they probably won't be able to provide since they will be shifting all the time and adjusted to up engagement and lower non desired outcomes etc.
3 comments
[ 612 ms ] story [ 1052 ms ] threadIt may seem pedantic, but the root of my question is that when I read the article, I don't actually understand what it is they are going to do.
(Also, from what I understand, there is lots of manual adjustment going on. So if they publish an algorithm for how tweets are ordered, or weighted, that would be before subjective manual adjustment anyway, making it at partly meaningless, no?)
I’m anticipating being disappointed.