Jitter is not a defense against statistical analysis. Additional lag might be what you want.
Additional constant delay? Isn't that also easily defeated when you're measuring from multiple nodes?
You can only _add_ jitter. So presumably you make people think you're farther away than you really are. I guess if you're especially clever, you can manufacture responses to packets you haven't sent yet to throw people off.
Or just block inbound ICMP.
No. You would need to add hundreds of milliseconds depending on the sender's location.