We show that some network and pom-pom constitutive models are essentially the same. Instead of the usual confrontation, we suggest that the two approaches can offer useful mutual support: vital information about network destruction rates found from detailed pom-pom calculations can be used to improve the network models, while deductions about network creation rates can pinpoint areas needing further attention in the tube modelling area. A new form of the PTT model, the PTT-X model, results in improved shear and elongational flow descriptions, plus an improved recoil behaviour. The remaining problems of strain-time separation, second normal stress difference description, and reduction of parameters are also discussed and some suggestions for progress are offered.