For a "normal" company under "normal" circumstances, the one year graph of the implied volatility is a pretty reliable scenario of how it will behave in the near future.
Well, sure you can... as SLE has elaborated; the var-swap is simply a replication. You can use tea-leaves if that's your bag. Price a MSFT var-swap against the atm vols and you'll see the convergence. This works under the assumption that the distro of future vol is normal. Under a large vol of vol[skew analog -- strike vols reflecting a large vol or vol] and convexity-bias it becomes trickier, but convergence is still seen to atm vols, with a slight premium in index var-swap/atm implieds.