Now if you want to believe that systems that rely on technology are better than using than the old 'dog and bone', respectfully, I disagree.
It sounds like the dog and bone beats the current computer system because the current computer system isn't very good (my understanding is that an operator, who might not understand or be able to find out the constraints of a station can log a request which might, or might not be possible to satisfy, possibly using ambiguous language, which then gets faxed(!)/e-mailed to the station, which is then used to organise assitance requests).
Theoretically, a more joined up system could be much better. It could help optimise the route choices (don't change at X, change cross-platform at Y), and do useful things like validation to make sure requests are readable and possible (and give meaningful reasons if they aren't), help with the allocation of assistance jobs at large stations, new requests could be added and quickly allocated, no-shows or extra information could be reported by staff at initial station/guards to help those further down the line. There could be a decent feedback system so any impossible requests are reviewed and the validation system changed.
But I do agree, the strength of people with local knowledge and being able to just directly contact the person at the other end, shouldn't be underestimated. A computer system needs to be designed to complement and make use of this knowledge.