The problem is that there's nothing inherent about a hypothetical that elides irrelevant details, it elides only whatever the author of the hypothetical wants t

The problem is that there's nothing inherent about a hypothetical that elides irrelevant details, it elides only whatever the author of the hypothetical wants to elide. Hypotheticals can let us agree on one aspect of a situation, but it can't make us agree on how relevant that aspect is to the situation we're actually interested in. This is why the trolley problem as applied to autonomous driving is so problematic, whilst it's an interesting thought experiment, its totally irrelevant to the person actually designing a self-driving car. It's a perfectly valid position to take that a situation is complex enough that hypotheticals aren't ever going to be helpful in reasoning about it.