This is another thing that client-side hit detection can help with. By offloading the ballistics calculations to the client, server performance will be improved just by having less stuff it has to do. It's a win-win for both improving general gameplay performance and helping out those servers that run beyond their capabilities. It's unquestionably an easier solution to implement than trying to split tasks up into concurrent threads without breaking the engine.
100% client side detection is bad. The server will still have to do some work to verify that the client is sending sane data. How much of a reduction in load that would be, I have no idea. How effective are anti-cheat measures like PB and VAC preventing cheating from lying clients, I have no idea.. (and it'd be a constant war between cheat-makers and anti-cheat app devs anyway.. always is)
While I can appreciate the frustration high ping players are experiencing right now, I'd be very interested to see empirical test using the current system... that could tell you things like, does playing on a under-populated server improve your clients perception? (if so, then it's likely the 'server too busy' issue because people are trying to get away with to much proc power they don't have)..
Can you find a "high-tick-rate" server.. any improvement? etc.
I like data.. I just do. Not that I'm in a position to do anything about it. .but still.. useful data helps people solve problems correctly. Anecdotal evidence, does not.
All of that said, Oz's internet does suck (I'm not bashing Aussies, I'm just stating a fact that they have comparability poor quantity of under sea cables, geographic issues, the Government likes to muck with the ISPs there requiring DPI, etc.. etc..), so I can understand the pain of persons stuck with using it for real-time applications.