Perhaps we got it wrong for exactly the same reason that using such minimum settings is a bad idea in many cases. SQL Server will hold the sum total of MIN setting idle if those pools with the MIN setting aren't being utilized. Therefore, if you have ten pools (pool1 through pool10) that HAVE 10% MIN and 100% MAX, and all pools are sitting idle until pool4 gets a big load, 90% of your resources will remain idle while pool4 struggles to handle the load with only 10% of the available resources.

Yes, the example is extreme, but so is the waste made possible by (mis-?)using this feature. The rationale is that a long query could tie up the resources for a while if allowed to utilize another pool's MIN amount, so it must sit idle when that pool isn't using it.