Whether 1MB is ideal or not, it's what we have.  It's pretty evident that Bitcoin would be lucky to avoid annihilation on attempts to change it in any way at this point in time.  That may or may not be the case in the future.

As it happens, 1MB seemed to have been at least quite fortuitous for us, and I wonder if it were not somewhat well considered when Satoshi made the setting as opposed to the perception promulgated by some that he pulled a random number out his ass.

 - It is/was just under the realistic limits needed to run the network behind TOR.

 - It got us all the way into mid 2015 before it became much of a stressor at all.  Had it been smaller, interest in Bitcoin may have died while at the current interest levels it is more likely that significant efforts will be made to solve any necessary engineering efforts rather than to walk away.

 - Even now that touching transaction rate limitations is within sight, the transaction body is chalk full of freeloaders who do nothing worthwhile for network support (e.g., Multibitch users) and there is much room to grow by simply allowing these users to drain away as a transaction fee market develops.

 - It looks like for some time the entire blockchain will fit on a MicroSD meaning that new nodes can be brought up with nodes which are modest in expense and a concealable bit of physical media.  There is an easily calculable and reasonable ceiling on the power and bandwidth necessary to operate this support infrastructure, and it is low enough in value that most operators can walk away from their investment if there were pressures applied from the mainstream legal and law enforcement world.

In retrospect, 1MB seems like a pretty ideal setting for the past history of Bitcoin and some distance into the future.  To me.