My domain is more bioinformatics than GIS, but the way I imagined it was that if one was arguing that [thing] data is better, they’re arguing that if more people recognised the innate benefits of [thing], we wouldn’t have to rely on software that uses [other thing] so much, and that to properly utilise [thing], it would take a bit of radical reworking of workflows, but there would be significant long term net benefit.
Basically, I think arguments like this tend to be more grounded in the socio-cultural practices of a research field than the absolute technical merits of an approach. Like in my domain, a DNA sequence is just a long sequence of 4 different letters (A, T, G & C), but there’s a bunch of ways we can encode that data into a file, many of which have trade-offs (and some of which are just an artifact of how things used to be done)