Peter, Keith and David (me) looked for a quick filler while we were waiting for another game - although it ended up taking the full hour.
I've been playing the PC version a fair amount, but the AI (even on the middling level!) still crushes me. However, it has taught me a fair amount about how to adapt your play to what other people are doing; as I have finally learned, in a VP game it doesn't matter what your final score is, as long as you are at least one point ahead of everyone else. And because St.P is essentially an open game, this is more possible than usual.
As I got first dibs on the Orange cards*, I obviously opted for the "keep 18 cash in case the Mistress shows up" tactic. This also has the advantage in the building phase of allowing you to pass once and see what the others build and/or take. Keith took an Observatory and then probably cost himself the game by using it to consistently draw from the Upgrades pile. This can be a decent strategy, but it still seems to me that taking workers for at least a couple of rounds has to be better.
I think you need to take a building into hand during that first round; it's useful to be able to avoid fighting another player for a particular category as the multiple discounts are significant, but you need to declare an interest in something too.
*The orange cards are clearly not Nobles (that's what the players are); it seems more intuitive to consider them as the white-collar workers you hire as compared to the green cards who are clearly the blue-collar workers.
As it happens, it was a Judge who appeared in the first line-up, so I bought him and kept my fingers crossed. And then I hit really lucky. I took a Weaving Mill from the Upgrades offering, despite not having a Shepherd, but first pick on the second round of workers gave me a Shepherd without giving either of the other players a fourth worker - that strikes me as being particularly significant. The extra income was probably enough to push the game totally out of their reach and the rest of the game consisted of my buying pretty well everything I wanted and the others taking increasingly desperate gambles to try and catch up, largely to no avail.
The final scores were: Keith 92, Peter 105, David 146 (which is by far the highest I have ever scored in a three-player game), and I very nearly won without the orange bonus at all!
I don't personally consider that the orange bonus is the most significant aspect of the game. It has a large impact, certainly, but even the ten point gap at the top end should be counterable with a well-focused building strategy and at least some attempt to keep up. No, my main problem is with the entirely arbitrary nature of the Upgrade cards. A game in which one player happens to get a couple of key green upgrades in the first two rounds, followed by the better orange upgrades is going to be hard to stop, and the very last round of upgrades can prove to be a bit of a scattershot too, given that green and blue upgrades don't count but orange ones may.