Let's hope I didn't miss some stupid little thing. Can't wait for DS's response...
I don't really see how Ducks could possibly fail. They have lots of margin to please Saladin and cap Mehmed in order to beat us. Edit: if they experience a small problem, they can just leave UN 1t from completion, instead of having to wait for a second vote. That gives them 32t to get Saladin's vote and cap Mehmed.
The biggest single improvement we could have made would be to build the Oracle in the capital instead of Bombay. That way we could have built the Pyramids, H.G, and forge in Bombay. This would have guaranteed us an engineer, as well as allowing us to generate an additional GP.
Probably right. Got me thinking: yes, it would have been much better. Generally, I would think that a necessary GE (for UN or Mining-powered space) is much better in the second or third city, with a capital dedicated to max research. At the time, it seemed like the best option - maybe a delay in Pyramids would've been ok?
We sacrificed potential scientists in Delhi for some 100t, about half of that with Oxford, and also didn't build GLib with marble. With just 2-3 scientists more on average (doable, I think), that's approximately 5k beakers (or 5t at the end). Probably more with GLib. Another GS bulb is 3k (3t), and an extra GP (maybe doubtful?

) would be another 2-3t. That's potentially as much as 11t off research. We were estimating an 8t UN slow-build/rush-buy. That's -6t compared to GE. So, 5t faster. If we got a low-odds GE and an extra GP, we'd be 8t better, which is their likely finish date.
We did have the option on capturing Aksum and another city, like LC suggested. That's clearly a good move in retrospect. We could've even done it very close to the end to good effect.
We're only losing by about 7t/250t, though, and to a team with a ridiculous roster. That's hardly a bad result. I'm more than happy.
Edit: We got Sci Meth 8t faster than them. They pulled ahead partly due to 2GG (6t worth of research) and the larger empire in golden age, I guess. Still, that's 28t research for us, to their 13t in the end.
