The difference between good plays and Payloads thus seems to be VP gaining. You know that Butcher is great for milling Provinces and getting them, so you get it as a Payload. Maybe there's another game where Butcher isn't great, but suddenly, you decide you need one for non-VP reasons. It's not a payload if it's not getting you points. I would define payload as "Potential for point gaining".
This. Payload is how you make up for the fact that you have been buying stuff instead of points, so it is either in the form of produce high amounts of points in a short amount of time, which is I guess the more classical idea of an engine, or in the form of be able to no doubt buy a high-value victory card each turn, in which case the engine is more about speed versus the money strategy.
In the latter case, the engine's payload is economy and is usually the result of strong trashing and/or a strong non-junking attack while lacking +buy.
The former case is more what the descriptions above are about, and is more variable. Obviously there is the classic economy and +buy, but stuff like remodel variants or goons can make an appearance. In either case, this is the one where the three pile endings tend to happen quite a bit more often because there is the potential for all the gaining, and this also tends to require a much larger engine, the creation of which depletes piles to begin with.
The engine on this board is of the second type, though I think that the best money strategy which looks something like witch/soothsayer/1-2 counterfeit/2-3 wharves here is being somewhat underrated. With good luck, it can get a vp lockout by around turn 17, and based on my games against bots (yeah, okay) on average around turn 20 if the margrave is not coming in hard early with the curse split usually 5-5 or 6-4 in its favor. Playing against my engines (which are not the greatest), it was probably about 50-50, with most of those games being able to tell what the result was going to be by about turn 10.
Having said that, an engine strategy that does not buy soothsayer will lose to it even if it gets a margrave in every turn by about turn 12, which is unusual with a lookout/silver opening. There is just not going to be enough power to make up 4-5 provinces once it gets going. Soothsayer/remodel is the way in which you make up this point deficit, sometimes even to the point that you want two of each. Then again, this board is more about ordering of 5-costs.
Assuming a lookout/silver opening, with villages being bought on 3 and 4 unless one gets far behind thinning, in which case second lookout.
Regardless of what the opponent is doing, the first one is witch unless it is turn 8, in which case see below and you are probably already screwed.
Against money, the margrave pretty much needs to be bought by turn 10 if you don't want a game coming down to the wire - late margraves tend to help the money strategy more than hurt it, so getting them up early is key. If lucky, it is your third 5 buy after counterfeit, if not, it is your second. Against an opposing engine, I am fairly sure wharf over margrave unless you are hitting 5 for the third time on turn 6ish (again, early hurts more), and even then maybe not. Once again, the second five is counterfeit.
First soothsayer once the deck is getting about 4/5 drawn. At this point it is usually not drawing the opponent a card
First remodel once the deck is getting all the way drawn. This is actually your only way to target-trash stuff and turns lookout(s) into either wharf or second soothsayer.
If needed to catch up on points, second remodel.
At least that is how I would go about the engine. Possibly slow against in a mirror, but in that case nowhere near as many points have to be scored, so the second soothsayer or remodel is probably unnecessary. A lookout/silver opening might be wrong since it can have issues hitting 5, and I think these issues are where a lot of the strength of the money lies.