I used to travel quite regularly from Lancaster to Manchester during the early 1990s and again around the turn of the year 2000.
My memory of the early 90s trips is a bit hazy, although I do recall loco-hauled Mark IIs on what became Cross Country used to run from the WCML to Castlefield via Bolton ultimately replaced by Voyagers and then transferred to TPE. On Saturdays I seem to think some of the local trains southbound from Preston went to Victoria and some to Piccadilly.
But by the 1999 to 2002 period, it was already dire. Everyone and their dog was routed via Castlefield. Going home from Oxford Road, it would be easier to get a train back to Pic and board there since otherwise it would be standing room only to around Horwich / Lostock. And platform 13/14 was then, as now, just horrible. An on time departure was very much the exception. So I couldn't understand when they proposed Ordsall Curve what on earth they hoped to achieve, and I still can't understand it now to be honest! Even if Castlefield were four tracked, and Piccadilly had two new platforms (or 10 new platforms), you still end up with the dogs dinner of flat crossings to the west.
I can't see what two new platform faces would do to help. The problem, as has been repeated ad nauseum above, is the decision that all services should go to Pic and the airport.
Network Rail's paper option C makes some effort to reduce that, but it's still not radical enough.
This is a fantasy thread of course, so we can propose any ideas.
1. Move Metrolink (partially) across to the Deansgate GNR viaduct to the West. The Eastbound CLC line can then use space on the Cornbrook viaduct to cross over the Castlefield lines and then join a bit further on, grade separating them.
2. Likewise at the Ordsall junction for the Windsor link, grade separate Windsor northbound traffic with a diveunder the Eccles line.
The above two removes some of the conflicting moves. Nothing you can do about Ordsall chord (other than go back in time and not build it, but we are where we are).
3. Extend the platform 1 at Oxford Road. It looks like there is space on the formation to do this up to a bridge at Gloucester Street, but this could be extended over the road relatively simply.
4. Put lifts, etc., into platform 1.
This gives you the same platforming capacity you would have had with 15 and 16 to allow trains to wait if necessary. It also allows platform 1 to be used as a terminating and turnaround platform if necessary without blocking the rest during the turnaround time.
5. Chat Moss stoppers, Southport / Kirkby, Blackpool and Barrow / Windermere go to Victoria, or the odd one can go to Oxford Road (perhaps one path an hour can be alternated between them). Additional peaks go to Oxford Road only.
6. There's space to slew the Metrolink lines at Victoria to the south east and therefore shift the line entering platform 1. Platforms 1 and 2 could therefore be extended under Cheetham Hill Road.
7. There's also plenty of space including the old chord from Victoria to Queen's Road for turnback sidings so that Eastbound terminators don't hold up the platforms.
8. If push comes to shove there's also space for a "Victoria West" terminating couple of bay platforms for peak hours.
For shoppers / many commuters / day trippers, Victoria and Oxford Road are more convenient than Piccadilly anyway. So why the obsession with everything having to run through to 13/14 and on to the airport?
Onward connectivity is one, but there are plenty of ways to interconnect without all services having to serve Pic. It's a legacy of 1980s rationalisation. Like when they wanted to close Marylebone station in London.