From my viewpoint, GraphQL is yet another great battle in the never-ending saga between the "full stack" crowd and the "client vs server" crowd. This battle ha

From my viewpoint, GraphQL is yet another great battle in the never-ending saga between the "full stack" crowd and the "client vs server" crowd. This battle has been raging since WAY before Web 2.0 and SPA -- at least as far back as Susan in accounting sneaking in a DOS PC running Visicalc and Word Perfect instead of using the mainframe. It is so seductive to imagine that "opening up the back end" is the answer. But, as many here have pointed out, unfettered access to the data model and transactions leads (yes) inevitably to the kinds of ad-hoc interactions which no back-end can rapidly adapt to. Kind of. It doesn't rapdily adapt mainly because the server team doesn't want to be dominated in the organization by the whims of Marketing and the client team. This leads client team to demand fuller access, because it is tired of its creativity and velocity getting dragged down by the slow roll of the server team. Throw in an occasional economic downturn with pressure from management to rein in back-end costs and this becomes a drama truly worthy of popcorn-munching.