When discussing the relative merits of programming languages, people speak of the language and its ecosystem interchangeably. For good reason: the two are not separable; with few exceptions, you can’t mix-and-match language and ecosystem.
Ecosystems are important. Things like HTTP servers, JSON/YAML/etc. libraries, database access and migration tools, etc. are commodities. If you’re building a commercial product, and development regularly halts for two weeks while the team builds some infrastructural component that is widely available in the ecosystems of every major programming language, you should reconsider your priorities.
But what is rarely appreciated is that the importance of the ecosystem is not constant with time.
Software projects have two phases: in the early phase, development is frenetic, and you’re adding new libraries to the dependency manifest on a near-daily basis as large volumes of the codebase spring out of the aether. Projects that are not abandoned eventually reach the late phase, a stage of maturity where work mainly consists of maintenance and refinement of existing components, and rarely are new things added to the dependency manifest. Features are still being added, but these features are typically more of the same (e.g.: a new endpoint, a new database access object), which is to say they use the existing dependencies.
In the early phase, the ecosystem matters most. Adding a line to a dependency manifest is more conducive to a flow state than implementing an entire library yourself. Getting to a demoable product earliest matters.
In the late phase, where the role of the ecosystem is fixed, the language matters most. When the ecosystem is all used up, the intrinsic properties of the language become visible. The reliability of software, the ease with which it can be maintained or refactored, the ease with which new people can be onboarded: these all stem from the intrinsic properties of the language, not the ecosystem. No amount of libraries is going to help your productivity if the highly-dynamic nature of the language makes confident refactoring impossible.
This is why Dropbox, Instagram, and Facebook have expended huge resources on enabling typechecking of their Python codebases, and why Facebook created a typed PHP.