Great post Ben. I like the ‘holding on to both with both hands’ as you work through the various constraints, conflicts, the actual problems and outcomes you’re really after.
I find it’s safer and clearer to use different words to describe these concepts, even though the approach is the same. That is so that we distinguish user needs — functional or emotional - and to protect the use of those words.
And then to talk about what it is that needs to happen to support or deliver a service, using neutral language. Such as checking eligibility or verifying identity — that doesn’t pre-suppose whether it’s a person or a computer doing that. And to talk about the underlying goals, the intent and outcomes for policy and the operational provider(s) of the service, rather than using a similar word ‘government needs’ to represent all of this.
Just as you say, ‘prioritising one set of needs over the other doesn’t work’. If we use similar words for both it feels a little easier for people to conflate the needs of the provider with the needs of the user, whether on purpose or accidentally. Even though the hard work is as you say, to balance the both, to make services that people prefer to use, that achieve policy intent, that are efficient to run, and that are easier to change in future as we realise how wrong or right we were.