> Google seem to be guessing at people's intended search query rather than performing the query based on the actual terms. I've read that google uses "Machine

> Google seem to be guessing at people's intended search query rather than performing the query based on the actual terms. I've read that google uses "Machine Learning" for their search results which I interpret to be exactly as you say they provide a stereotyped result based on what they think you are searching for (possibly optimized either for what is inevitably clicked on or for ad revenue), instead of actually matching terms. What this means is that search results may be more accurate in some statistical way, like more people click on the top result, but it also pumps up the number of edge cases where it guesses wrong, while simultaneously making it impossible to tell where the results went wrong because you can't understand how they were generated (compared to eg keywors search where good or bad, the reason you got a result is obvious)