I think people argue about which language is better for much the same reasons they argue about which mobile phone operating system is better, or which console, or which text editor. They've invested (in the case of programming languages, potentially many years) in something, and want to feel like they've invested wisely. I think that's also probably just human nature.
I disagree with the notion that programming languages need to be popular, at least, as popular as they needed to be in the past.
20 years ago it was important for languages to be popular because that was the only way they'd reach enough critical mass to be discoverable. Growing up, there was only one book store within driving distance of my house that had any programming books. That meant if you wanted to learn programming, you were limited to what had been "blessed" by industry (in my case, one book on C++ that was actually just a syntax reference). That seems to no longer be true (the internet has made discovery of programming languages easier in much the same way it's made the discovery of everything easier).
There is certainly a benefit to having critical mass, but that mass seems to be much smaller than has ever been required in the past. I've seen Haskell developers on several occasions talk about the fact that Haskell's popularity has been in this nice "happy medium" that provides enough eyeballs to flush out issues and provide feedback, but also allows them to not worry about breaking things.
A language getting popular is not all benefits with no drawbacks. It gets harder to make breaking changes to the language (Javascript can be considered a good cautionary tale about what can happen when a language becomes so popular that you are held hostage by backwards compatibility). And you can't read a single article on here about some less popular language that doesn't prompt at least one comment about how because the language isn't popular, the quality of developers using it is higher (which is always posited as a plus for hiring developers in that language). Paul Graham mentioned it regarding Python explicitly in one of his essays.
More people being part of a community means more everything. More of what's good (libraries, eyes looking at bugs, ideas on solving problems) and more of what's bad (bike-shedding, shitty libraries, conflicting programming idioms).
I have no idea if there's some magical sweet spot, where you get all the benefits of a language being popular without any of the detriments; but I suspect it doesn't exist. As it stands, yes, I'd like there to be more Haskell libraries. I'd like there to be more blog posts about using Haskell for things that aren't abstract math. But I'm not necessarily itching for Haskell to become much more popular.
We're already starting to see some of the industry people that use Haskell start to explicitly target evangelism (like the work FP Complete is doing). I think it's great that they want other people to use this thing that they like, but I am worried anytime a community switches from "stealth" mode to "evangelism" mode. Incentives around addressing problems starts to become perverse when you have to worry about messaging, and when you're trying to convert other people. It seems like it evolves eventually to cargo-culting.