You've probably heard, or even said, how bad Bing's search results are in quality. It's true, they were pretty bad. Keyword "were". Back closer to the launch of Bing (then MSN search), the index size was significantly smaller than that of Google. However, here's the thing... it's been 21 years since it launched, and it has improved slowly over time a lot since then, and I have the data to prove it!
I was curious about the quality of different search indexes and their respective query engines, so I decided to run a very small experiment. I set up a simple survey of sorts where participants were given results for various search queries of 6 different search engines, except all branding was stripped so they'd have no idea which engine it came from. They were then told to rate the quality of the results for each engine in terms of how useful they'd find it. Despite only making 2 Reddit posts asking for people to take it and the posts appearing to not receive much attention, they actually got 206 individual engine-query-quality data points and the results converged enough to be somewhat reliable. So here's what I found.
As you can probably guess by the title and the entire first paragraph of this article, Microsoft Bing was a strong leader. On a scale from 0 to 5, where 0 is absolutely useless results and 5 is incredibly useful, Bing scored the highest with an average of 3.65, securing its place as the most useful search engine statistically. In second place came Qwant at 3.56, which partially sources its results from the EUSP, and as you might guess, partially also from Bing. Third place went to DuckDuckGo quite a bit lower at 3.43, which, shocker, partially uses Bing results as well (alongside results from its own crawler and specific niche sites)!
At last, we get to the engines that are not powered by Bing. The first non-Bing-derived engine is actually Brave Search at 3.35, which is fully independent from any other index. They have a bit of an advantage in that their browser has an opt-in feature to save any website you visit to their index, which lets them find commonly used sites in methods besides crawling quite efficiently.
Only after all these 4 previous engines do we get Google at 3.32, as a very close runner-up to Brave. Let this be a lesson to you that being the biggest player doesn't mean you're the best player, as they have insane amounts of resources, compute, and finance, yet they are second-to-last here. Finally, we get Mojeek, with its own independent index, sitting at 3, sadly relatively quite far behind all the others.
If you want to do the survey yourself to see how it worked or contribute to further improving the data, feel free to do so at survey.nilch.org, it is completely anonymous and none of the data you provide will be tied back to you. I think that this small experiment proves just how irrelevant compute can be to quality of search results after a while, showing law of diminishing returns apply very strongly to search.