isn't a weak board part of the point? I wanted to address solely the saboteur issue, not whether squire/gardens is good or not. Your claim was that saboteur can be a counter to it, so I must assume that it means games where the engine alone is not strong enough to beat gardens anyway, otherwise it's a worthless thing to claim.
You can have loan, would that be sufficient?
You are completely right that this is a marginal question. I know that a strong engine that gets unlucky, e.g. a T5 Chapel, can claw back a win after Squire/Gardens gets enough bloat to beat all the provinces; what I don't know is exactly where the line is drawn. Certainly Sab on the board increases the engine's marginal win rate where you would beat the rush but lose to the slog.
I'll give it a go with Loan, would you be free tomorrow night (EDT) to try? I'll be free then, but then I'm off to the beach on vacation.
Now, there are several problems here. Firstly, KZ said "probably," which means he was never wrong in the first place, no matter how fast Transmute or Squire is. Secondly, you made a lot of arguments why Squire into Gold is better, but KZ never said it's not better, he just said it's not faster, which probably means "how many turns does it take you to gain your first gold via Transmute/Squire->Trash->Trash for benefit". That's a very different thing. And also, he was using Transmute as an example, as a method of describing the speed of Squire, he never actually cared about how fast Transmute is, so even if what he said was wrong, it's not really a big deal.
The problem with this is that the speed, thus definied doesn't matter. Take Silver vs Transmute.
Transmute:
T1/T2: Potion/x
T2/T3: Transmute/x
Now we have 14 cards, on average we get 1 estate a hand. It is 3 hands to get through the deck so we expect to hit Transmute/Estate on T6.
Silver:
T1/T2: Silver/Silver
T2/T3: We expect to have an estate in each hand so we expect to hit 5's so we buy silver/silver.
T5: We have 4 silvers, 7 coppers, 3 Estates. We have $8 from Silver and $7 from coppers. That gives us $15 and hence an average hand of $5. Now we do have some high variance so we should expect one hand in T5-T7 to have 2 Silvers, 2 Coppers.
In other words, by this metric Transmute is no slower than Silver.
What about other cards? Chapel/Market Square is one of the fastest, most reliable shots at gold.
T1/T2: Chapel/Mrksqr
T3/T4: You have a 10/12 chance chapel is in one of these two hands naturally; if it is you have a roughly 4/11 shot that Mrksqr is in the same hand. 30% Chance of a T3/T4 gold.
T5: We dropped 4 cards, likely bought a second Mrksqr. Odds are maybe 2/3rds that you hit your gold here.
Can you do faster still? Sure Death cart is an easy Gold. Baron has good odds for gold quick, MrkSqr/Smithy/Mint(or Doc) ... but those sort of things are the absolute fastest you get and all have their own problems.
Generally, when people say Transmute is slow, it isn't that you get your first Gold on T6-T7, it is that it
slows down your deck and that getting more stuff later is even slower. If speed to first gold is what he means, then again, it goes back to his sense of how things work not translating well into syllogistic statements; Transmute is a pretty much worthless benchmark for the speed to get to gold; it is about as fast as the average card (certainly faster than stuff like Spy, Hag, Oasis, Taxman, Mine, Market, Witch, Lab, or Hunting party).
The ultimate question is how
how useful is Squire's on-trash setup. About a quarter of potential Squire boards have at least the theoretical ability. A subset of those will actually be somewhat viable. More often Squire/TfB/$5 will make some better option more reliable. I had a T5 Chapel that hit CCCC. I can still use Squire for Minions/Goons/Pillage/Jester/etc. I only talk about fast golds insofar as that one of the things that is easy to calc shows that Squire can be competitive. For things that are harder to calc, Squire is a lot better.
Develop/Squire/good draw attack/decent 4 is a thing. You can develop Squires into drawing attacks. You can Develop Estates into Develops, Develops into $4/Squire and Squire into $3/drawing attack. Is this faster than raw Develop? Certainly. You don't have to spend 5's on strong drawing. You can use them on payload (like Merchant Ship) or for more reliability (like Lab or Cartographer) or for accelerants (like Butcher); you can also run more Develops as Squire/Develop/Develop/C/C isn't a bad hand.
And I can go on with lots of highly degenerate setups. None of that is going to show up in just 3000 games worth of sampling. Even it it halves Zog's loss rate; he'd have to have a fairly low win rate for his experience to conclusively show one way or another that Squire's on trash benefit is not that useful. Looking for the less common Squire/TfB/$5 is going to take even yet higher data. This is why I don't like looking at single player's experiences - a good player with a bias towards or against certain strats will have their own skill swamp the effect of the strategy.
I just don't know what to tell you. Should I let drop that I play Dominion against a professional statistician from the BLS? Should I let drop my 10,000 games (though a LOT of those are pre-DA and a huge number have been where we played semi-random kingdoms)? I certainly can remember where Squire/scaling TfB/Attack has won me the game, but would I have won without it? Zog clearly thinks not, I think likely. The numbers certainly are not there to determine it. It is pretty much a worthless show stopper to say "I haven't seen X be useful in Y games" unless Y is abysmally large or your win rate is very close to parity.