The difference is not 8ms, it's closer to 4-6ms.
If you play with people on a Toronto server, the latency is identical (probably better for TSI due to not going there over the public internet).
If you connect to a server in Montreal, that latency would indeed double, because the Montreal server needs the data to get to Toronto. Possibly worse, because the public internet path isn't as fast as Bell's internal network.
In the real world, I can give you some figures:
Home -> Toronto Server: 13ms
Home -> Montreal Videotron Fibre: 22ms
So on VDSL2, you could reasonably ping 22ms to a Montreal server. But anything else an it becomes less clear; routing through Toronto to Chicago, for example, adds no latency (again, probably faster than the public internet).
Besides this, there is no discernible difference to the gaming experience because of an added 9ms of latency. It's the placebo effect. You're talking about half a frame of latency at 60FPS, and all modern games have sophisticated latency correction anyhow. Far more important than the difference in latency is the quality of the latency. Jitter and packetloss. The same is true for VoIP. Even then, games have excellent extrapolation and interpolation to account for these.
EDIT: Also, realize that if you press the "fire" button on your computer, and your round-trip latency is 22ms, then it takes roughly HALF that time for your computer to send the notification to the server. Because 22ms is there-and-back, but sending commands to the server is just one-way. And there's no reaction-time loop here either, because your local game is predicting the enemy's location anyhow.--
Developer: Tomato/MLPPP, Linux/MLPPP, etc »fixppp.org