Would the problem work if the student started summing backwards from the last item? How many items would the student have to swap in that case?. For how many random combinations of items would the problem "work" (i.e. require exactly one change?)
On the other hand, in your example you're not just trying random things until something works. Noticing that 1 + 2 = 3, etc is more analytical than adding members at random and hoping that the person who created the problem thought of the same random pattern you thought about.
The dollars-and-cents addition problem mapped onto the "fun" toy store narrative is not mathematically interesting until assigned the type of analytical firepower outlined in OP's blog post. It is a deliberately obtuse trick question posed as a learning tool; children labelled "gifted" will figure it out and feel the dull satisfaction of killing another "problem", and those without that benefit will simply experience more fear and boredom.
"Smart" children are identified early. After that they are smart because everyone knows they are smart. It is very hard for a person who knows they are not smart to convince a society who knows they are not smart.
Dumb people are sometimes lucky and get something right. That doesn't make them smart. Smart people sometimes make mistakes. That doesn't re-label them as dumb. The important thing is the label you start with. Smart people will recognize the grammatical "mistake" in the previous sentence and understand that this post came from a dumb person.
For example, consider this problem for geometric series: 1 + 2 + 4 + 8 + 16 + 32 + ... + 4096 = ?. This problem may look completely intractible and tedious without a calculator. But once you start adding: 1+2=3, 1+2+4=7, 1+2+4+8=15, 1+2+4+8=31, ..., even a child can spot the pattern. Now that they are amazed, they will be motivated to learn more about the cause of this pattern, etc, etc.