Re:The Vivaldi browser will not support Google’s FLoC
by throwaway3699 on Tuesday April 13, 2021 @05:25PM
(#26799531)
Why would that need ML? PageRank has worked for ages for this problem.
Re:The Vivaldi browser will not support Google’s FLoC
by teraflop on Tuesday April 13, 2021 @11:21PM
(#26802578)
I'm not sure what you think PageRank is, but it doesn't produce links between documents as output; it requires them as input.
Re:The Vivaldi browser will not support Google’s FLoC
by sweetdreamerit on Wednesday April 14, 2021 @02:42AM
(#26803738)
I would consider PageRank as a quite successful example of machine learning. Am I wrong?
Re:The Vivaldi browser will not support Google’s FLoC
by xnx on Sunday April 18, 2021 @02:47PM
(#26855157)
The definition of what counts as ML/AI can be fuzzy, but I don't think the original PageRank algorithm would qualify as ML/AI under any definition: http://infolab.stanford.edu/~backrub/google.html
Re:The Vivaldi browser will not support Google’s FLoC
by warlog on Wednesday April 14, 2021 @06:21AM
(#26805090)
It started out that way, anyway.
Re:The Vivaldi browser will not support Google’s FLoC
by dang on Wednesday April 14, 2021 @01:17PM
(#26810087)
The goal with those 'related' lists is to get something more meaningful and interesting than HN Search results (which are pretty good already). That's a high enough bar that it takes judgment and taste to know what to include.
For sure we want to get beyond "mods do it manually" as the strategy. I'm not averse to trying out an ML/AI approach—the nice thing is that all the relevant data is public, so anyone could build it and we could all try it out. But my personal inclination is to build something that would rely more on the community.
(Maybe even a "HN search found 3 possibly related threads")