In general, people are trained by Google to expect a lot out of a search. On Google you can phrase a search in a number of ways that don't precisely map to how the ideal content phrases it, and still figure it out. This works because Google has a massive amount of data - data from searches about what words are generally synonyms of other words, data from other pages on the internet that link to the target page, and data about what past searchers have done.
When you do a site-specific search, or a search on an internal corporate site, much of this data is often totally absent. So it may just not be possible to build a search with as high a quality as you expect, on that corpus, with modern technology.
Source: I don't know anything about Reddit search specifically but I was formerly a search expert and I am pretty sure this is what's going on here.
I don't know, I rolled my own search feature for a previous project just using mysql full text search with weighted results and ended up with something that worked way better than reddit's search function. Building the next google is hard but building a functional search feature isn't exactly rocket science, especially when you have reddit level money.
I personally think the search sucks because good search takes a lot of resources (in many different ways) and they prefer to allocate those resources elsewhere.
I remember creating a search function in school, and while it worked pretty well, I'm not sure it would work well for an organization with as much information as Reddit.
As for reddit-level money, does reddit even make all that much compared to their expenses?
Is that a fair comparison? Reddit has a ridiculous amount of data to search through. I feel like if improving search was so trivial, they would have done it already.
Well they do have a search that works reasonably fast, so they have the infrastructure already. It's just not that good. It's a algorithm/methodology issue. I can understand if they were still in startup mode or it wasn't 2017. I'm not saying the method I mentioned would be a solution for something on reddit's scale but we live in an age of readily available data analytic solutions and open-source search engines. An open-source solution like Solr would work, I just don't think reddit has any motivation to solve the issue.
Edit: Found this on reddit's wikipedia page:
"On July 21, 2010, Reddit outsourced the Reddit search engine to Flaptor, who used its search product IndexTank.[91] As of July 12, 2012, Reddit uses Amazon CloudSearch."
It seems to be an implementation issue, not a technological one.
I'm guessing reddit search simply isn't used often enough for them to dedicate resources to improving it. The site is primarily about having a constant stream of new posts anyway. Worst case you can use Google.
It is provided by YC Company Algolia. Many YC companies leverage the software provided by other YC companies.
If you were a one of the first YC companies and you're core business was to aggregate, curate and influence the development of content. And you weren't able to develop a core service, hypothetically something like search. Maybe you could buy that service from some other company; like Algolia.
Potentially, it would work the same way as HN does. The site they cloned when they pivoted.
When you do a site-specific search, or a search on an internal corporate site, much of this data is often totally absent. So it may just not be possible to build a search with as high a quality as you expect, on that corpus, with modern technology.
Source: I don't know anything about Reddit search specifically but I was formerly a search expert and I am pretty sure this is what's going on here.