Josh. Individuals working in the “scatter/gather” model will _always_ be less efficient. They form individual bottlenecks as people wait for them. They create unnecessary integration steps and a failed integration adds considerable dev time. Since they don’t have external expertise available as they work, they spend time looking stuff up and flailing around. They cannot benefit from the learning inherent in working collaboratively with people who have skills they don’t have. Their code requires an extra review step. Their tests are typically inadequate, so there’s an extra testing phase. I could go on. These issues can easily double or triple development costs. The question, then, is whether the company is willing to eat those additional costs. It’s a simple cost/benefit calculation. Sure, people can refuse to work collaboratively because they don’t feel comfortable with it. That’s equivalent to a carpenter refusing to use power tools because they don’t feel comfortable around them. Yes, they can get the work done, but there’s a cost.
I should add that if there are “intense social interactions” in an Ensemble, you’re probably going about it wrong. You cannot work in an ensemble without psychological safety.
In this whole discourse about "ensemble" programming you are having - you still haven't addressed the issue of individual suitability for such an approach!? Some people find the intense social interactions involved to be actively damaging to their mental health...