I disagree. Being able to quickly understand what kind of range something will be in is a very important skill. I've met far too many engineers and engineering

I disagree. Being able to quickly understand what kind of range something will be in is a very important skill. I've met far too many engineers and engineering students who have no real concept of magnitude. In an academic setting, that can lead to comically large or small answers. In real life it can lead to lots of wasted time. Asking someone to talk through a problem like those discussed is often very useful. It gives you an insight into how they think, the background knowledge they can draw on and their understanding of accuracy of data. It's fairly obvious when someone's bullshitting (especially if they talked through the process with you), and any good candidate will be honest about the inaccuracies inherent in their assumption.