Comment by simonw
2 days ago
An interesting intellectual exercise is to think about how a search engine could provide the best possible answer (from a user satisfaction perspective) to a query like "best CBD gummies".
A lot of people have a significant financial incentive to win at that search query.
What would the perfect top search result for that look like?
It would probably be an article written by professional writers in a trustworthy publication with a strong ethics policy, provably followed over the years, concerning whether they accept payment for promoting specific products in supposedly impartial reviews.
If you can figure out how to algorithmically detect that kind of content you could build a pretty great search engine!
I think I'd be pretty happy if Consumer Reports was on the top for queries like these (if they had the relevant data, of course). I think they follow your criteria pretty closely.
Since "the best" doesn't exist, just like there is no magical professional that has unique insight into the mind of the user making the search, a search engine could become pretty great by simply not taking decades to remove scams like the one described in the article from the top of search results
There are many criteria for "best" that are acceptable to many people, e.g. lowest price, proven high quality ingredients, efficacy, etc. "Website with high reputation that happens to be running ads for company XYZ" is usually not how people define "best".
By splitting one very ambiguous term into another "high quality" one you haven't resolved anything.
And people do transfer trust from the medium to the product, so neither is the second filter very robust when it comes to people perception
And the notion of "the best" is inherently subjective
Are we assuming that this search engine is only used by a few nerds, or is the idea to build something that remains good even if it gets popular enough that webmasters have strong financial incentives to game it like they currently do with Google Search? Because the latter sounds like a much, much harder problem, and in particular like it probably requires huge financial resources in order to win the ongoing cat-and-mouse game, if that's even possible.
For the sake of this exercise I want to hear how people would solve that harder problem.
I think it'd maybe a query for the best gummies would be based on reviews from users, but I guess that's the point. Having something understand what one means by best is hard.
Hm, I think that Amazon shows that just user votes might not be sufficient - e.g. because users can be paid off to give 5 star reviews, which bias the results.