Yes, but that's because the space contains so many duplicate or almost duplicate articles that search results have no chance of being good. I see the problem in the searched content, not in the search engine itself.
The internet also contains "so mamy duplicate or almost duplicate" articles, and yet Google can provide a useful search for Internet articles. You're telling me it's somehow impossible to achieve the same thing for a company's intra - even though it has several orders of magnitude fewer articles to index?
One quick and easy way to improve Confluence's search would be to rank often-visited articles higher than rarely-visited articles.
I would hope that going by popularity is something a search engine WILL ABSOLUTELY NOT DO in a corporate setting. This is a completely different scenario than what we have on the public WWW, so unless your corporation has a situation where people maliciously copy articles from others, trying to pass them off as their own content, in which case the search engine quality is the least of your problems. :)
Anyway, as I've already stated, Confluence spaces which are well looked after and are not allowed to get cluttered are a delight to search in, and as a bonus also a delight to just browse through.
P.S. Google Search results are often lousy and full of low quality results anyway, I'm not sure mentioning it makes your point as well as you think it does. As everywhere - crap goes in, crap comes out. :)
Personally I haven't seen "well looked after" Confluence spaces. I have experience with 4 different companies that used Confluence for their infra, and in my experience it has always been a mess. The reason I advocated for ranking by popularity is that those messy Confluence pages are filled to the brim with obsolete outdated articles that nobody intentionally visits. For example, in many firms people post meeting notes to Confluence. So you have 10 000 pages of quality "notes from project abc weekly meeting 5.4.2014". Now, imagine yourself searching a Confluence space like that. You type in a particular phrase, and there is a perfect match in meeting notes from 2017, a perfect match from preliminary design document from 2013, etc. These results constantly drown out the actual pages you are searching for, which are typically more actively-updated, more recent pages. It would be a huge improvement to rank these pages higher purely based on their popularity.
