Memory constrained systems like phones and tablets may well not want to waste 10MB on getting better diagnostics - especially in the production case where the exact kernel version should be well known to the manufacturer/service provider - so a crash backtrace can be processed offline to reconstruct file/line information.
For development on larger systems (laptops/desktops) ... I'd like to remind people that it is 2012 and RAM costs less than $0.005 per megabyte. So "wasting" 10MB in this way costs less than a nickel. So if you take Linus's suggestion of firing up gdb to help decode things, and you only value your time at baseline USA minimum wage - you'd need to complete your analysis in 25 seconds to come out ahead. And that's assuming that you only look at one stacktrace in the lifetime of the system.