It's only with the last few expansions that strategy has tipped so heavily toward engines.
It's hard to disagree more 
The best card for engines overall, Chapel, is from base. Base also featured some A-grade village (plain Village) and an A-grade draw card (Smithy).
The worst attack to get against you every single turn, Ghost Ship, is from Seaside.
The reason engines were harder with Base only was that engines require a bit of everything. +actions, +cards, +gain/buy, time (either attack or alt VP).
In the base set a lot of cards provided only one of those categories, so the chance that a category was entirely missing from a kingdom was significant. It was still easily below 50% though. Engines were probably also harder with base + intrigue, again because of the high percentage of cards doing only one thing. But cards stopped doing only one thing way, way before Dark Ages.
The real reason engines are so much more popular now, is that everybody got a lot better at Dominion.
I'm convinced WW builds an engine now on many many boards he used to think BigMoney was best on. And he could think BigMoney was best on those boards back in the days because he almost always beat his opponent with it, who was either playing a worse form of BigMoney or terribly misplaying an engine.
I agree that there's far more to my play improving than the cardset changing, though I think they're both factors. I disagree with some of the things though. Certainly I don't think engines are great in base in general; I imagine it can't be right much more than 50% of the time. There's only Chapel as a really good trasher; moneylender also existed. 35% of the time already, you don't have either of these. It's not impossible to engine without either, but it's tough. Witch certainly wrecks engines that don't have Chapel. And then you need a Village (there are only 3, Festival is pretty bad for engine reliability, Throne Room often isn't great since there aren't many cantrips), you need draw (Smithy is great, Library and Lab are ok but not the best), you need +buy (Woodcutter is bad, Festival is ok, Market is good of course)... most given boards will have any one of the given pieces, but usually not all of them.
Perhaps the bigger thing, surprisingly enough, is that the payload is rather poor. There are very few options that get you off of getting a decent amount of treasure, at which point how much good are you really doing over BM? So of course Chapel went engine a good majority of the time, but other than that, it wasn't happening that often.
Even getting Intrigue out (should have) made a big difference, as you pick up Steward, Masquerade, Torturer, Bridge... Your biggest problem now was Villages, because you only pick up Mining, Nobles, and the very-unreliable-especially-against-money Tribute. If we discount Nobles as being real bad as the only village for an engine (and same of course with Tribute), we have 4 villages out of 50 cards, which means we get one about 60% of the time (Throne Room sometimes doesn't work, but I suppose sometimes Nobles does).
Seaside was out before all of these discussions, and for engines it brought Ambassador, Lookout, Wharf, Fishing Village, Bazaar, Native Village... Should have been pretty engine already.
But the thing is, things kept getting better for engines, too (Alchemy; Prosperity did a LOT; Dark Ages added very good villages and several very good trashers, and attacks...), in no small part because the proportion of each component kept creeping up, making it much less likely you'd miss something you need. Still, it's not a foregone conclusion that you engine every board - just grabbed the last 20 games I played off of salvager, and I think 7 of them are pretty clear no-engines (with a couple being questionable). That's obviously not scientific and a pretty small sample, but it wouldn't surprise me at all if engine isn't right ~30% of the time.
A final note is that there are way fewer do-nothing cards now. Spy, Scout, Thief, Chancellor, etc. all being so weak hurt engines significantly, as they effectively shrink kingdoms. The engine % would be way higher if we played with 12 card kingdoms, and lower if we played with 6.
