docker has a lot of semantics that make it much more lightweight than a virtual machine. This makes it very nice for the type of throw-away exploration that @snkfischer is talking about (or microservices, etc.). It also does not have its own kernel (it uses the host machine's kernel), which makes images much less resource-intensive, starts up quicker, etc.
On a typical workstation, a few VMs will cripple the host machine's resources because you are running several operating systems concurrently (and lots of overhead with it). With Docker, it's way easier to run a whole bunch of containers concurrently without resource issues, and it pretends to have a separate operating system, so you still have some really helpful isolation. There are some nuances to this, because this does mean it is more tightly coupled to your OS than a VM will be, but that's what the community here, the Docker community, and documentation are for! 