The Hardware Recommendations community is pretty unique among the Stack Exchange communities. You've pointed out a big reason: questions are very niche. I think they are even more niche than Software Recommendations. Anecdotally, I have searched Software Recommendations for software that meets my needs several times, but that someone else asked about. Here, not so much. Hardware and Software are two different beasts.
That isn't bad though. We are certainly helping users. However, we have a large number of unanswered questions and we are going to reach parity of questions and answers very soon. We are posting multiple answers to questions infrequently, and I think this is a larger problem than the placing on hold that occurs (more on that below). With only a single answer, we aren't providing users with a set of options. We're basically saying "here, get this" instead of "here are a handful of options and here are the differences between each". With a single recommendation, the users are not getting the full benefit of our experiences and knowledge. Multiple possible recommendations can point out benefits and draw backs to other hardware, provide cheaper/more expensive alternatives, or just provide a better endorsement of a particular brand of product (perhaps, one has better customer support). Questions with outdated answers aren't closed. We can provide newer recommendations if a better product is released. We should be doing this.
The question of placing on hold has come up many times, as I'm sure you are aware. Many of these are due to either the broad nature of the question (what components do I need to play a game?) or the technical support-y nature of the question (my hard drive is clicking, how do I fix it?). Failure to close these would make the site even more confusing. According to the close stats, we are reopening very few of these questions. In the last 90 days, we haven't reopened any technical support questions (and only 7 out of 114 were even edited). For the broad ones, we've reopened a couple. Overall though, your perception appears to be correct, we've closed 57% of new questions in the last 90 days. There is a problem with defining our scope (or at least what users perceive our scope to be). Suggestions on how to fix/improve that should be in another meta question though. It's going to be a topic that requires a lot of discussion.
As I mentioned above though, I think the bigger problem is the question to answer ratio. Area 51 says we are at about 1.5 answers per question, but SEDE says it's closer to 1.2 and has been almost since the beginning.
Sorry about the colors, those are assigned by SEDE
The question/answer ratio is the light blue line. It's been hovering at 1.2 since about the third week of the site's existence. At the same time, we generate at least two comments per post (likely clarification questions) and we are voting less on both questions and answers. So, we have a site that involves a time commitment on both the asker and the answerer (clarification questions, possible edits) and a smaller and smaller reward for that work. Honestly, after seeing that graph, I was surprised how steady the answer ration has been, despite the falling vote lines. That's a testament to our dedicated users providing answers.
My SQL skills are very bad (I didn't write the above query, it came from an excellent meta post on Code Review), but I would be interested in seeing the voting ratios on our posts that have more than one answer. With that, I'd also be interested in seeing the accept rate on one answer versus multiple answer questions. I hypothesize that the accept rate is slightly higher, but the number of up votes per answer is higher than one posts with only a single answer.
This messy graph (again, sorry, I took it from the same post above) shows some other activity stats on Hardware Recommendations.
What I see in there, though, is that we aren't rewarding our good questions and answers with up votes any longer.
Back to your original question: What is the community good for and is it doing it's job?
When we have an on topic question, we are good at finding something that meets the user's criteria. The key here is "on topic". We are very particular about what we want - which component, what price range, etc. - which makes it hard for users on the first try.
Are we doing our job well? I'd put us at a "slightly below average". Our scope issue continues to plague us. We receive a lot of technical support questions that just don't belong here. We aren't voting as much, which is seen by the distribution of reputation. We are a year and a half old and only have 4 users with more than 2,000 rep. Zero users are at the "Trust User" (4000) reputation level. From a community point of view, that puts a lot of the burden on the moderators. We aren't self sufficient yet.
From the point of view of a user asking a question, we can appear hostile if a question isn't asked perfectly. With some back and forth and edits to the question though, we do provide users with answers. In this case, we are doing our jobs, for that user. No one has 5K rep yet to access the analytics, so I'll need a mod to answer this: What does the site analytics say about our traffic from Google?
How do we improve?
- When we engage with a user to get more information, that information needs to be edited into the question. It's much easier when someone else comes along to read the question with all the details than reading the question and then a short thread of clarification questions.
- Vote up the good content and vote down the bad. Voting in both directions is important.
- Try to provide a second opinion on answers. Give users another option.