One of the biggest downsides of the Haskell ecosystem IMO is the lack of good libraries. Or rather, well-documented libraries and I am not talking about docstrings here. Most library authors will write some comments and let the document generator and type system do the work for them. It's not how it works and Haskell will never take off until this issue has been tackled. Most people will have a hard time figuring out how to make a real-world project, while people who are not beginners will keep hitting brick wall after brick wall until they either reach some sort of God-like understanding of Haskell or give up entirely and rewrite the project in another language.
This is not entirely on the library authors either. For some reason the entire Haskell community is adverse to writing good documentation and/or tutorials. Everyone seems to be enamored with the idea that the terminology used should be as cryptic and unaccessible as possible and that everyone should be able to completely grasp the type system and all the various intricacies that one finds in a language like Haskell since day one. Now before someone starts thinking about the reasons why this is, I'd like to mention that I know very well about Haskell's start in academia and how deeply rooted in maths it is, but unfortunately most software engineers and most importantly project managers do not care about function purity or how you can rewrite crucial parts of an infrastructure by using co-monoids and functors. What matters are results, ease of development, and a strong knowledge base that unfortunately is lacking.