WTH - automations are not doing plausibility checks?

When I set up an automation and have a condition that checks whether a certain entity is switched off to only act when it is switched off, there is little sense in preselecting “off” as an action for the very same entity.

Similarly, when I have an if-then-construct, it is very unlikely that the then clause should also switch off an entity, if the if-clause is already doing that.

Also, when I select devices with multiple states, I tend to see the least useful state to be preselected quite often.
E.g., thermostats should default to temperature selection, as this is probably the most likely state one wants to alter.