The PageRank algo by definition is broken, because it equates popularity rather than truth or correctness.
The perfect example of why this is broken is health and medicine, for all its fails, we all know that modern medicine is based on evidence, something all other alternatives are not.
Secondly take news, the actual source the source of truth most of the time is not the first match, and could and is often surpassed by some social media commentator which is just wrong imho.
PR never worked.
Its the same as thinking the stars of music (pick any name) are the best because they are popular even if they cant play an instrument and most never write their own songs.
Correctness is impossible to evaluate without a panel of humans to define "the truth" for each topic and an advanced NLP system that compares the meaning of a page with it. Impossible in 1996, you could maybe pull this off today with LLMs but it would be super expensive and not very reliable.
This would create an intentional bias in the results. That's not something you expect from a web search, it should return the most relevant and popular links not the ones that agree with mainstream science. Wikipedia, PubMed and other databases where actual experts contribute are way better for that.
Comments
The PageRank algo by definition is broken, because it equates popularity rather than truth or correctness.
The perfect example of why this is broken is health and medicine, for all its fails, we all know that modern medicine is based on evidence, something all other alternatives are not.
Secondly take news, the actual source the source of truth most of the time is not the first match, and could and is often surpassed by some social media commentator which is just wrong imho.
PR never worked.
Its the same as thinking the stars of music (pick any name) are the best because they are popular even if they cant play an instrument and most never write their own songs.
Correctness is impossible to evaluate without a panel of humans to define "the truth" for each topic and an advanced NLP system that compares the meaning of a page with it. Impossible in 1996, you could maybe pull this off today with LLMs but it would be super expensive and not very reliable.
This would create an intentional bias in the results. That's not something you expect from a web search, it should return the most relevant and popular links not the ones that agree with mainstream science. Wikipedia, PubMed and other databases where actual experts contribute are way better for that.
You clearly don't understand how Page Rank works. It is not just a matter of counting the incoming links.