A business model is scalable when revenue (or sometimes profit) is a non linear function of inputs, which are basically people and capital. In
r = f(p, c), if f is a non linear function, then the business model is scalable.
In the InShorts example, it is true that if you double the number of people, you'll double the content generated. That's linear. But if by doubling the content, if you get four times the visitors and four times the revenue, that is non linear and scalable. Similarly, if FreeCharge has to spend x amount to acquire a customer, and repeat rate is low, then revenue will just be a function of x. That model will not be scalable.
Network effects,, virality, automation, self service, aggregation etc. are tools to build scalability into your business model. While some business models are inherently more scalable than others, it is the actual design of the model and how you incorporate these tools that really matters.
A couple of other items to consider:
Time to scale. The function f mentioned above can acquire many forms. In some cases, revenue might grow rapidly earlier and then slow down. In other cases (especially SaaS), you might see a long gestation period after which revenue starts growing exponentially
Resource limits. However well your model is designed, if availability of resources is limited, you will reach a limit. For instance if your model requires Tulu-speaking PHDs who can solve the Rubik's cube in 2 minutes while moon-walking, you are unlikely to be able to scale beyond a certain point.