Why Subnet you private network

[rubberneck]

If you are designing a network from scratch why would you bother sub dividing your network? I know subnetting is for security and to help broadcast traffic however - for example, why not just choose a private class B range with the default subnet mask say: -

172.24.x.x
255.255.0.0

giving you a possible 65,000+ hosts (you dont have to user them all)

Then if you was concerned about traffic on that segment or you did want to subdivide for security (say for another department) - why not just choose another class B. Lets say: -

172.25.x.x
255.255.0.0

then use a router between the two as you would have to if you subnetted.

Why would i bother say giving that 172.24.x.x network a subnet mask of something like: -

172.24.x.x
255.255.252.0

which would then cut down the amount of hosts to 1000+ and give 64 extra networks (each with 1000+ hosts)

As the class B private IP range is from 172.16.x.x to 172.31.x.x what is stopping me from just saying one segment is on

172.24.x.x
255.255.0.0

and one is on

172.25.x.x
255.255.0.0

I could do this for 16 segments on my network and never have to worry about not having enough host IDs.

When you look at subnetting tutorials and examples they seem to always start off saying "you are given the IP range of . . . .". So i am assuming that is from the likes of an ISP etc. In those cases i see the point of subdividing your given range.

I suppose my question boils down to - Why not just use another one of the private IP ranges instead of subnetting up one of them.

Thanks.