Re: Baldwin vs Alco. I've read a few things that make me think that Alco was a bit ahead of Baldwin when it came to calculating driver balancing. The balancing was a very complicated engineering problem, because you had TWO things you wanted to balance: rotating mass (so as not to get the hammer-blow effect Eliphaz describes), but also reciprocating mass: the piston, piston rod, and driving rod are going back and forth, and their momentum would tend to make the locomotive yaw back and forth. Trouble is, you can't balance both: the balancing weights you need to do one tend to make the other worse.
Now, again this is my impression from what I've read: I am not a trained mechanical engineer. But one of the advances in locomotive design around 1930 was the introduction of one-piece, cast, locomotive beds in place of the earlier built-up frames. What this meant was that the problem with reciprocating masses became less important: the locomotive structure was more rigid, so could resist the forces better, and also more sturdy, so less likely to get bent out of shape or loosened by the effect. So Alco's tendency was, I think, not to try as hard to balance the reciprocating masses: late Alco designs for the New York Central still had some counterweighting positioned to counteract the effects of the reciprocating parts, but they didn't try to compensate for as large a proportion of this effect as earlier designs had. So the weighting could be placed to do a better job of compensating for the rotating masses, and so do a better job of eliminating hammer-blow. New York Central Hudsons (an Alco design) and Niagaras (another Alco design) were smooth-running locomotives. ACL 4-8-4 and New Haven 4-6-4 (Baldwin designs) had problems when first delivered. (I guess that the fixes involved moving counterweights-- often these involved, I think, lead weights put in "pockets" in the steel driving wheel-- so as to approximate better to the sort of balancing that Alco got right the first time!)
As I say, this is the impression I have gotten from reading about a number of locomotive classes. It could be all wrong… or it could be that the "drive train dynamicists" (I'm sure that wasn't an official job description, but I think it describes the relevant part of the engineering task) at Schenectady in the late 1930s were better at it than their counterparts at Eddystone.
Leading to a nice bit of counterfactual history. Suppose Baldwin and Alco's bean-counting had come out a bit differently, so that Alco got the New Haven order. It's time to work out the details. Alco's engineers tell the New Haven's "I know you haven't used boosters in the past, but they would let you get a high tractive force when starting up with smaller, lighter, pistons in the main cylinders, and I don't think I have to tell you what that would mean for operations and maintenance. And if you agree to that, we can shave the price a bit, because we can use some available casting designs and not have to come up with something new…" And the New Haven people see the light and agree. Result: an Alco I-5 with different enough streamlining sheet metal to disguise it, but mechanically a copy of the New York Central's J-3.