There is also a lot of value in a question that works well enough, that you ask consistently over long stretches of time (or long stretches of distance). Maybe it's not perfect, but the longitudinal data would be worthless if they updated the wording every single year.
Although I'm no survey expert, the thing I'd like to bring to everyone's attention is how easy it is to not take into account people that have a degree of numeric or math illiteracy... which I guess they are the main target demographic that is included by these questions (and I can also guess that they make a worryingly large part of the demographic, because our systems are rarely inclusive).
In my experience, having met people from multiple countries during the time I've been living abroad, what I have noticed is that — in this world filled with inequality — it is a privilege to be able to have a good grasp in scientific subjects. And, for lots of different factors, people have setbacks or trauma that make it difficult to learn a subject that is either boring or painful to them.
So, yes the questions are a bit convoluted, but they help paint a mental image for probably the majority with a thing that they may be closely familiar with: stairs... Plus, it probably helps statisticians get a better signal to noise out of the questions, too.
Adding this nuance to the question serves to invite deeper thought and avoid assigning a motivation-based rating (like when you give the Uber driver 5 stars when what you felt was actually just "satisfactory").
A more basic rating question can invite other kinds of influence, such as a motivation in how they'd like their life to be perceived rather than how they genuinely feel it to be.
In surveys with less nuance the data tends to correlate around the extremes.
I'd also bet that they found the above "convoluted" question was one that led to the same people giving more consistent answers from day to day and moment to moment.
Even if I'm wrong I hope you see this is a much thornier problem than just asking a question and assuming the answer tells us anything about the person taking the survey.