One Pentium® III 600MHz Processor and one Celeron® 700MHz processor, which is better

Interesting analogy. I'd say that the CS equivalent of the kind of house-building you're describing involves preferring the use of OTS components to custom stuff -- e.g., if writing C++ code, use STL vectors rather than writing one's own expandable-array clbutt. It seems to me that knowing when to take which approach (OTS stuff versus custom-built) is critical. Maybe with houses custom-built is always better if one can afford it, but I'm not convinced this is true in software. (Though it might be.)

But what you say about computing still doesn't make clear to me how *knowing physics* adds significant value for people who aren't going to design hardware. You say

but I would claim that you don't have to know much if anything *about physics* to "know the bits".

Is it maybe that without physics the physical limitations of what hardware can do don't make sense? e.g., if you know no physics, you don't understand that there's a limit to how fast information can move from one place to another?

To try an example, consider two people working in operating system design-programming. One of them understands how processors work in terms of circuits ultimately constructed from switches, where the switches are implemented using transistors, but he-she has no clue how a transistor works, below the level of "as a switch". The other also understands how a transistor works. How does this additional knowledge help him-her do a better job of o-s design or programming?

(I'm not necessarily arguing with the person I'm replying to here, just trying to steer the discussion back toward my question.)

-- B. L. Mbuttingill ObDisclaimer: I don't speak for my employers; they return the favor. -- -- blm