I said it will take more time, not the same time.

Yes, I'd scan the range (in some amounts of distributed different subranges, in some whatever order), with 100% confidence of finding the key along the way, at some point.

Your way means having less than 100% confidence after going through some ranges, and when things don't work as expected, then what? Rinse and repeat. More effort to build up to 100%, yes. And a gazillion intervals starting off from weird offsets. Keeping track of what was scanned and what not. Everyone understood already for a long time what you are doing, you don't really need to explain so many times.

But you do you.